U.S. patent application number 16/363676 was filed with the patent office on 2019-07-18 for system and method for feature recognition and document searching based on feature recognition.
This patent application is currently assigned to OPEN TEXT SA ULC. The applicant listed for this patent is OPEN TEXT SA ULC. Invention is credited to Simon Dominic Copsey.
Application Number | 20190220481 16/363676 |
Document ID | / |
Family ID | 52132867 |
Filed Date | 2019-07-18 |
United States Patent
Application |
20190220481 |
Kind Code |
A1 |
Copsey; Simon Dominic |
July 18, 2019 |
SYSTEM AND METHOD FOR FEATURE RECOGNITION AND DOCUMENT SEARCHING
BASED ON FEATURE RECOGNITION
Abstract
A system for document searching can include a camera. The system
may further include an image capturing module configured to capture
a first image of a first portion of a document, a feature
recognition module in communication with the image capturing
module, the feature recognition module configured to determine a
first feature associated with the first image, a search module
configured to send search information to a server and receive a
first result from a first search of a set of documents that was
performed based on one or more search criteria determined based on
the first feature associated with the first image, and a search
interface configured to present the first result on the device.
Inventors: |
Copsey; Simon Dominic;
(Horsham, GB) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
OPEN TEXT SA ULC |
Halifax |
|
CA |
|
|
Assignee: |
OPEN TEXT SA ULC
|
Family ID: |
52132867 |
Appl. No.: |
16/363676 |
Filed: |
March 25, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16014528 |
Jun 21, 2018 |
10282374 |
|
|
16363676 |
|
|
|
|
15393806 |
Dec 29, 2016 |
10031924 |
|
|
16014528 |
|
|
|
|
15098923 |
Apr 14, 2016 |
9563690 |
|
|
15393806 |
|
|
|
|
14322591 |
Jul 2, 2014 |
9342533 |
|
|
15098923 |
|
|
|
|
61842138 |
Jul 2, 2013 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06K 2209/01 20130101;
G06F 16/285 20190101; G06K 9/00442 20130101; G06F 16/58 20190101;
G06K 9/228 20130101; G06F 16/5846 20190101; G06F 16/2455 20190101;
G06F 16/3344 20190101 |
International
Class: |
G06F 16/583 20060101
G06F016/583; G06F 16/33 20060101 G06F016/33; G06F 16/2455 20060101
G06F016/2455; G06F 16/28 20060101 G06F016/28; G06K 9/00 20060101
G06K009/00; G06F 16/58 20060101 G06F016/58 |
Claims
1. A method for document searching, comprising: on a client device,
capturing a plurality of optical character recognition (OCR) data
from image capture of a document on the device; deriving search
information based on a subset of the OCR data; sending the search
information to a server; receiving a result from a search of a set
of documents performed on the server; deriving enhanced search
information based another subset of the OCR data and the received
result; sending the enhanced search information to the server;
receiving an enhanced result from a search of the set of documents
performed on the server; and formatting the enhanced result as a
result output for display on the device and displaying the result
output along with the image capture of the document on the device.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of and claims a benefit
of priority from U.S. patent application Ser. No. 16/014,528, filed
Jun. 21, 2018, entitled "SYSTEM AND METHOD FOR FEATURE RECOGNITION
AND DOCUMENT SEARCHING BASED ON FEATURE RECOGNITION," which is a
continuation of and claims a benefit of priority from U.S. patent
application Ser. No. 15/393,806, filed Dec. 29, 2016, entitled
"SYSTEM AND METHOD FOR FEATURE RECOGNITION AND DOCUMENT SEARCHING
BASED ON FEATURE RECOGNITION," which is a continuation of and
claims a benefit of priority from U.S. patent application Ser. No.
15/098,923, filed Apr. 14, 2016, entitled "SYSTEM AND METHOD FOR
FEATURE RECOGNITION AND DOCUMENT SEARCHING BASED ON FEATURE
RECOGNITION," issued as U.S. Pat. No. 9,563,690, which is a
continuation of and claims a benefit of priority from U.S. patent
application Ser. No. 14/322,591, filed Jul. 2, 2014, entitled
"System and method for FEATURE RECOGNITION AND DOCUMENT SEARCHING
BASED ON FEATURE RECOGNITION," issued as U.S. Pat. No. 9,342,533,
which claims benefit of priority to U.S. Provisional Patent
Application Ser. No. 61/842,138, filed Jul. 2, 2013, entitled
"System and method for OPTICAL character RECOGNITION AND DOCUMENT
SEARCHING BASED ON OPTICAL character RECOGNITION," the entire
contents of which are hereby expressly incorporated by reference
for all purposes.
TECHNICAL FIELD
[0002] This disclosure relates generally to systems and methods for
performing optical character recognition and using the results of
such optical character recognition. Specifically, this disclosure
relates to systems and methods for performing optical character
recognition on a device and systems and methods for searching
documents based on optical character recognition.
BACKGROUND
[0003] While technology utilized today has led to widespread use of
electronic documents in certain environments, paper documents have
not been completely exorcised. As a result, in many environments
both electronic documents and paper documents may be utilized. The
simultaneous use of paper and electronic documents has imposed an
inconvenient and oft-clumsy segregation with respect to the
searching of such documents.
[0004] To illustrate, there may be a wide variety of ways in which
to search electronic documents. For example, in many cases
electronic documents may be managed by a content management system
which provides an interface that may be utilized to search those
documents (e.g., by keywords, content, etc.). Additionally,
electronic documents that may reside at multiple disparate
locations may be indexed and the index of those documents used to
search for a document. Thus, once a desired electronic document is
obtained (e.g., through searching or otherwise) the content of that
electronic document may be used to search other electronic
documents for other electronic documents that may be similar (e.g.,
with respect to terms contained, content, etc.).
[0005] Paper documents may also be searched. In many cases, paper
documents may be organized (alphabetized, catalogued according to
an organizational system, etc.) such that those paper documents may
be searched or otherwise located. Again, then, once a desired paper
document is obtained (e.g., through searching or otherwise), the
content of that paper document may be used to search other paper
documents that may be similar (e.g., through manual searching of
those paper documents).
[0006] In environments that utilize both paper documents and
electronic documents, it is oftentimes desired to find electronic
documents based on a paper documents. Currently, the only way to
perform such a search is to manually review a paper document and
use a provided search interface to manually perform a search of the
electronic documents based on the manual review of the paper
documents.
[0007] What is desired then, are systems and methods that allow
electronic documents to be searched based on paper documents. In
essence, a more seamless connection between physical paper
documents and the ability to search, find, and utilize
corresponding or associated electronic documents.
SUMMARY
[0008] A method for document searching, in accordance with
embodiments, includes capturing a first image of a first portion of
a document with a camera of a device; determining a first feature
associated with the first image; sending search information from
the device to a server; receiving a first result from a first
search of a set of documents that was performed based on one or
more search criteria determined based on the first feature
associated with the first image; and presenting the first result on
the device.
[0009] In some embodiments, the first feature comprises at least
one of text, document layout, or document formatting. In some
embodiments, determining the first feature associated with the
first image comprises performing optical character recognition
(OCR) on the first image to recognize text in the first image,
wherein the first feature comprises text. In some embodiments, the
search criteria comprise a search term formed from the recognized
text.
[0010] In some embodiments, the method includes forming a search
query comprising the search term at the device, sending the search
query to the server, the search information comprising the search
query and receiving the first result from the server in response to
the search query.
[0011] In some embodiments, the method includes sending the
recognized text to a server, the search information comprising the
recognized text; forming a search query for the first search at the
server based on the recognized text; performing the first search
according to the search query; and returning the first result from
the server to the device.
[0012] In some embodiments, the method includes capturing a set of
images of the document with the camera, wherein the first image is
captured as one of the set of images; determining a set of features
by performing OCR on the set of images, wherein the first feature
is one of the set of features; determining an overall feature
associated with the set of images, wherein determining the overall
feature comprises joining at least the first feature associated
with the first image of the set of images to a second feature
associated with a second image of the set of images; and
determining the one or more search criteria based on the overall
feature.
[0013] In some embodiments, sending search information from the
device to the server comprises sending the set of images from the
device to a content provisioning platform that performs the OCR on
the set of images and the determines the one or more search
criteria. In some embodiments, the performing of OCR on the set of
images and the determination of the one or more search criteria is
performed at the device and the first search is performed by a
content provisioning platform. In some embodiments, the overall
feature and the first and second feature comprise text features and
determining the first and second features and overall feature
comprises using natural language processing to identify text
features. In some embodiments, the method includes capturing a
second image of a second portion of the document and receiving a
second result from a second search of the set of documents based on
the second image of the second portion.
[0014] A system for document searching in accordance with
embodiments includes a camera; a processor; a computer readable
medium storing a set of computer instructions executable by the
processor to provide: an image capturing module configured to
capture a first image of a first portion of a document; a feature
recognition module in communication with the image capturing
module, the feature recognition module configured to determine a
first feature associated with the first image; a search module
configured to send search information to a server and receive a
first result from a first search of a set of documents that was
performed based on one or more search criteria determined based on
the first feature associated with the first image; and a search
interface configured to present the first result on the device.
[0015] In some embodiments, the first feature comprises at least
one of text, document layout, and document formatting. In some
embodiments, the feature recognition module comprises an OCR module
configured to recognize text, wherein the first feature comprises
recognized text. In some embodiments, the device comprises a search
criteria module configured to determine the one or more search
criteria from the first feature, wherein the one or more search
criteria comprise a search term and the search information
comprises the one or more search criteria.
[0016] In some embodiments, the system includes a content
provisioning platform coupled to the device. In some embodiments,
the search information comprises the recognized text and the
content provisioning platform is configured to perform searches and
return search results, wherein the first search is performed by the
content provisioning platform and the first results returned by the
content provisioning platform. In some embodiments, the device
comprises a search criteria module configured to determine the one
or more search criteria from the recognized text; the device is
configured to provide the one or more search criteria to the
content provisioning platform by sending the search information to
the server; and the content provisioning platform is configured to
perform searches and return search results, wherein the first
search is performed by the content provisioning platform and the
first results returned by the content provisioning platform.
[0017] In some embodiments, the first image is one of a set of
images of the document, each of the set of images is associated
with the document; the feature recognition module is configured to:
perform OCR on the set of images to recognize a set of features,
wherein the first image is one of the set of images and the first
feature is one of the set of features; and determine an overall
feature associated with the set of images, wherein determining the
overall feature comprises joining at least the first feature
associated with the first image of the set of images to a second
feature associated with a second image of the set of images.
[0018] In some embodiments, the device comprises a search criteria
module configured to determine the one or more search criteria
based on the overall feature, wherein the search information
comprises the one or more search criteria. In some embodiments, the
device is configured to provide the one or more search criteria to
the content provisioning platform by sending the search information
to the server and the content provisioning platform is configured
to perform the first search and return the first result to the
device. In some embodiments, the device is configured to provide
the overall feature to the content provisioning platform by sending
the search information to the server and the content provisioning
platform is configured to perform the first search and return the
first result to the device. In some embodiments, the overall
feature and the first and second feature comprise text features and
determining the first and second features and overall feature
comprises using natural language processing to identify textual
features. In some embodiments, the device is further configured for
capturing a second image of a second portion of the document and
receiving a second result from a second search of the set of
documents based on the second image of the second portion.
[0019] A method for performing OCR, in accordance with embodiments
includes receiving a set of images of a set of portions of a
document captured with a camera of a mobile device; determining a
set of texts by performing OCR on the set of images, wherein each
of the set of texts is associated with a corresponding image of the
set of images; determining an overall text associated with the set
of images, wherein determining the overall text comprises joining
at least a first text associated with a first image of the set of
images to a second text associated with a second image of the set
of images; and storing an OCR version of the document that includes
the overall text.
[0020] In some embodiments, the overall text is determined based on
overlapping text that is present in the first text and the second
text. In some embodiments, determining the overall text comprises
performing natural language processing on the first text and the
second text to join the first text and the second text. In some
embodiments, the method includes collecting motion data associated
with the set of images, wherein the overall text is determined
based on motion data associated with the first image and the second
image.
[0021] A system for OCR of documents in accordance with embodiments
includes a mobile device camera; a processor; a set of computer
executable instructions stored on a non-transitory computer
readable medium, the set of computer executable instructions
executable by the processor to perform a method including:
capturing a set of images of a document using the camera;
determining a set of texts by performing OCR on the set of images,
wherein each of the set of texts is associated with a corresponding
image of the set of images; determining an overall text associated
with the set of images, wherein determining the overall text
comprises joining at least a first text associated with a first
image of the set of images to a second text associated with a
second image of the set of images; and storing an OCR version of
the document containing the overall text.
[0022] In some embodiments, the overall text is generated based on
overlapping text that is present in the first text and the second
text. In some embodiments, generating the overall text comprises
performing natural language processing on the first text and the
second text. In some embodiments, the method further includes
capturing motion data associated with the set of images and wherein
the overall text is determined based on motion data associated with
the first image and the second image.
BRIEF DESCRIPTION OF THE DRAWINGS
[0023] The drawings accompanying and forming part of this
specification are included to depict certain aspects of the
invention. A clearer conception of the invention, and of the
components and operation of systems provided with the invention,
will become more readily apparent by referring to the exemplary,
and therefore nonlimiting, embodiments illustrated in the drawings,
wherein identical reference numerals designate the same components.
The invention may be better understood by reference to one or more
of these drawings in combination with the description presented
herein. It should be noted that the features illustrated in the
drawings are not necessarily drawn to scale.
[0024] FIG. 1 is a block diagram illustrating one embodiment of a
topology for document searching.
[0025] FIG. 2 is a block diagram illustrating one embodiment of a
topology for document searching.
[0026] FIG. 3 is a flow diagram illustrating one embodiment of a
method for document searching.
[0027] FIG. 4 is a block diagram illustrating one embodiment of a
topology for on-device optical character recognition (OCR).
[0028] FIG. 5 is a flow diagram illustrating one embodiment of a
method for on-device OCR.
[0029] FIG. 6 is a depiction of an example document.
[0030] FIG. 7 is a block diagram illustrating one embodiment of a
topology for on-device OCR.
[0031] FIG. 8 is a block diagram illustrating one embodiment of a
topology for back end OCR.
[0032] FIG. 9 is a block diagram illustrating one embodiment of a
topology for document searching.
DETAILED DESCRIPTION
[0033] Embodiments and the various features and advantageous
details thereof are explained more fully with reference to the
nonlimiting embodiments that are illustrated in the accompanying
drawings and detailed in the following description. Descriptions of
well-known starting materials, processing techniques, components
and equipment are omitted so as not to unnecessarily obscure
embodiments in detail. It should be understood, however, that the
detailed description and the specific examples, while indicating
preferred embodiments, are given by way of illustration only and
not by way of limitation. Various substitutions, modifications,
additions and/or rearrangements within the spirit and/or scope of
the underlying inventive concept will become apparent to those
skilled in the art from this disclosure.
[0034] Embodiments as disclosed may allow the searching of
electronic documents based on a paper document. In particular,
according to certain embodiments, an image of a paper document (or
a portion thereof) may be captured on a device. From the image, a
feature may be extracted. Such features may include, for example,
text, document layout, document formatting, a pattern of the spaces
between words or other feature.
[0035] For example, optical character recognition (OCR) may be
performed on the image to determine text of the image of the paper
document. Based on the determined text, a set of search terms may
be determined and a search performed on a set of electronic
documents using these search terms. The results of the search may
then be returned.
[0036] In some embodiments, a set of features is determined from a
set of images by using OCR. From these, an overall feature may be
determined. Determining such an overall feature may include joining
a feature from a first image with a feature from a second image.
The one or more search criteria may be determined based on the
resulting overall feature. The results of the search may then be
returned. In some embodiments, the features from the first and
second images are text features and the overall feature is
identified therefrom using natural language processing.
[0037] Depending on the embodiment, the OCR may be performed at a
user device that also captures the images. The device may then
determine the search criteria and send them to a content
provisioning platform to perform the searching. In other
embodiments, images are sent from the device to the content
provisioning platform, which then performs the OCR and
determination of the search criteria.
[0038] In some embodiments, a first search may be performed based
on a first image (or set of images) captured from a first portion
of a document and a second search may be performed based on a
second image (or set of images) of a second portion of the
document.
[0039] The results, which may be a set of documents responsive to
the search, may be returned in a variety of ways. For example, in
certain embodiments, the electronic documents found through the
search may be returned to the device on which the image was
captured and displayed in a list. Alternatively, a set of links to
the documents (e.g., in a content server or at the locations at
which the documents reside) may be returned and displayed to the
user of the device.
[0040] Moreover, in one embodiment, the results of the search may
be displayed in conjunction with the image of the paper document on
the device (e.g., as a list, set of links, initial sentence of each
electronic document, etc. overlaid on top of the image on the
screen at the user's device). Using embodiments such as these then,
an interactive search that may be substantially continuously
updated based on the image a user is currently viewing may be
implemented.
[0041] For example, in certain embodiments, as a user moves his
device over different portions of the paper document, and a portion
of the paper document is displayed to the user on the user's
device, a search is performed based upon the current portion of the
paper document being displayed on the user's device. The results
are displayed to the user in conjunction with that portion of the
document being currently displayed (e.g., overlaid on top of the
image of that portion of the document on the screen). Thus, as the
user moves his device over different portions of the document the
search results indicating electronic documents associated with
those different portions may be shown to the user and substantially
constantly updated as the user moves his device to those different
portions.
[0042] As may be realized, mobile computing devices (or just mobile
devices) such as cellular phones, smart phones, laptops, PDAs,
etc., are increasingly the way users conduct a wide variety of
their interactions. As such, many of the images used to perform
embodiments as presented herein may be captured and processed on a
mobile device. However, currently a number of impediments to doing
OCR on mobile devices exist, including, for example, relatively
inferior camera performance or sensors (e.g., poor low light
performance), the ability of the camera (or other image capture
device) on most mobile devices to capture images where the entire
image is in focus, the ability of current OCR technologies to deal
with images that are skewed or rotated and the large processing
power required to perform OCR on a captured image. Thus, currently,
performing OCR of a document on a mobile device may take an
inordinately long time or may fail altogether.
[0043] Despite these impediments, however, it is still desired to
perform image capture and OCR on or for mobile devices.
Accordingly, systems and methods to effectively implement OCR in
conjunction with a mobile device are presented herein.
Specifically, according to certain embodiments, a set of images of
a document may be captured, where each image may be of a portion of
the document. The portion of the document may be a portion of a
page such that the image does not capture the entire page. OCR may
be performed on each of these images and the text or other feature
resulting from performing OCR on each of the images may be joined
to form an overall text or feature corresponding to the images of
those portions of the document. Thus, a single overall text for a
document may be generated more efficiently by performing OCR on
images of a set of portions of the document and joining them
together.
[0044] In particular, by performing OCR on images that are
themselves only portions of a document, issues with respect to
lighting and focus of the image when performing such OCR may be
considerably reduced (e.g., the text of the images may be much
bigger). Moreover, the time to OCR images of portions of a document
would also be significantly less than performing OCR of an entire
document (for example, as a result of the reduction of the
aforementioned issues). Thus, performing OCR on a set of images for
portions of a document and joining them together to form an overall
text may be performed more quickly and fail less often, than
performing OCR on the same portions of the document captured as a
single image.
[0045] In one embodiment, when the text corresponding to each image
of each portion is obtained by OCR the resulting text may be joined
by performing language processing (e.g., natural language
processing) techniques to analyse the texts obtained from the
images to determine how they are to be joined. For example,
overlapping text in different images may be used to determine which
text should be joined and where. Additionally, in some embodiment,
in a similar manner to how predictive text on mobiles works, with
an appropriate language database on the device, text from the
images could be joined by evaluating probability of text from
different images matching.
[0046] Additionally, in certain embodiments, other data that may be
obtained from the mobile device may be used to inform the joining
process. For example, motion data associated with the set of images
such as the accelerometer or other motion sensor on the mobile
device may be used to determine a direction of movement of the
mobile device when the set of images were captured or orientation
of the device when the set of images were captured, etc. This
motion data may be used when joining the text obtained from the set
of images and may further improve both the results and the speed of
the joining process. Alternatively, or additionally, in certain
embodiments motion data such as movement between images or
direction of travel of the device over the document may be
determined algorithmically from the image data and the
algorithmically determined motion data used during the joining
process.
[0047] It may now be helpful here to discuss embodiments of various
topologies that may be utilized in performing embodiments as
disclosed herein. Referring first to FIG. 1, one embodiment of
topology for document searching is depicted. The topology includes
one or more computing devices 110 connected to a content
provisioning platform 120 over a network 130. The network 130 may
be a wired or wireless network such as the Internet, an intranet, a
LAN, a WAN, a cellular network, another type of network. It will be
understood that network 130 may be a combination of multiple
different kinds of wired or wireless networks.
[0048] Platform 120 may include one or more servers or other
computing devices providing content provisioning modules 122
accessible at one or more locations (e.g., IP addresses or domain
names) or through one or more interfaces. The modules of a
particular platform 120 may be deployed on physical computing
devices residing at a particular location (such as those associated
with the provider of a particular mobile application) or may be
deployed in a cloud. Thus, when a platform 120 is deployed in the
cloud, one or more content provisioning modules 122 may be
executing on a virtual machine provided in the cloud, where the
virtual machine is addressable at a single (or more)
location(s).
[0049] Regardless of the location of the platform 120, the content
provisioning module 122 of a platform 120 may support access from a
computing device 110. In other words, users at computing devices
110 may use their computing device 110 to access content
provisioning module 122 using for example, a browser or other
application on the computing device 110, a proprietary application
on computing device, a generic interface, etc. In response to such
access, content provisioning module 122 may provide data from data
store 121 to the accessing computing device 110. This data may
include documents 124, including for example, files in a
proprietary format (e.g., Adobe .pdf, Microsoft Word, Excel, Power
Point), files in a generic open format (e.g., mp3, mpeg, jpeg,
etc.) files in a markup language (XML, HTML etc.) or practically
any other type of file. Thus, for example, content provisioning
module 122 may be a content management system that provides access,
control and management of documents 124 in data store 121.
[0050] Accordingly, content provisioning module 122 may include
search module 126 including search interface 128 to allow for
searching of such documents 124. Device 110 may provide search
information to platform 120 for processing. Search interface 128
may be accessible (e.g., at a particular URL, through an API or web
services interface, etc.) such that a query including one or more
search terms may be provided through the search interface 128. The
search module 126 may search the documents 124 based on the query
provided through the search interface 126 to determine a set of
results responsive to the query, where these results may be
associated with a set of documents responsive to the query. These
results can then be returned by the search interface 128 in
response to the query.
[0051] Computing devices 110 may be mobile devices (such as
smartphones, laptop computers, personal data assistants (PDAs),
etc.), desktop computers, servers, or other computing platforms, or
any other type of device that can process instructions and connect
to network 130. Each computing device may include an image capture
module 134 and a search module 112.
[0052] Image capture module 134 may have access to a lens or other
aperture on a device configured to receive or capture images using
the light received through the aperture. For example, many mobile
computing devices include a camera lens or the like and an
associated image capture capability which the image capture module
134 can leverage. Thus, image capture module 134 may, when
utilized, receive image data (with or without direct user
involvement) through the aperture of the device 110 and capture or
otherwise provide these images.
[0053] Search module 112 may include interface 113, feature
recognition module 114 and search term module 116. In some
embodiments, the feature recognition module may be an OCR module.
Interface 113 may have access to one or more input or output
modules (e.g., a screen, buttons, speakers, etc.) on device 110 and
be configured to provide output or receive input through these
input or output modules of the device 110. Interface 113 may also
be configured to utilize other interfaces such as APIs, web service
interfaces, etc. to send requests and receive responses to those
requests.
[0054] Specifically, in one embodiment, interface 113 may be
configured to be accessed by a user such that the image currently
being received by image capture module 134 is displayed on the
screen of the mobile computer device 110. In one embodiment, then,
interface 113 may allow a user to indicate that an image (e.g., the
image currently being received by the image capture module) is to
be used for a search when activated by a user (e.g., when the user
selects a particular button or touches a screen in a particular
place, etc.). Interface 113 may also be configured to present the
results of a search to a user (e.g., in conjunction with an image
being displayed to the user or in another type of interface
altogether).
[0055] The feature recognition module 114 may be configured to
identify features of an image. For example, in some embodiments,
the feature recognition module may perform OCR on an image, where
the result of the OCR of an image is text corresponding to that
image. In one embodiment, for example, such an OCR module may be an
ABBYY OCR module or the like. In other embodiments, features
extracted or recognized can be or include font types or document or
image characteristics such as amounts or configurations of
whitespace or patterns formed by space between words or other
features on a document. In one embodiment, for example, a feature
recognition module 114 can be configured to identify the blank
areas between words and create of a "fingerprint" for that
pattern.
[0056] Search term module 116 may be configured to determine one or
more search terms from the output of the feature recognition
module. In the case of text, for example, search term module 116
may include a natural language processor (NLP) module 118 or the
like configured to remove stop words from the text, determine key
words or phrases, performing key word or term weighting, etc. Thus,
search term module 116 may determine a set of search terms from
text by, for example, using all the text (e.g., each term of the
text is a search term, or the entire text is one search term),
using a set of key words or phrases determined form the text, using
the words of the text after removal of stop words or by determining
the set of search terms in some other manner.
[0057] Search module 112 is thus configured to be activated and to
access image capture module 134 to obtain images (e.g., image data)
from image capture module 134 and provide these images to interface
113 to be presented to the user. Search module 112 is also
configured to receive an indication that a user wishes to use an
image currently being displayed for a search. The search module 112
is thus configured to provide the image to feature recognition
module 114, receive the text associated with the image from the
feature recognition module 114, provide the text to search term
module 116, receive the search terms from search term module 116,
use interface 113 to provide the search terms to search module 126,
receive the results of the search from interface 113 and present
the results of the search using interface 113.
[0058] Accordingly, a user of computing device 110 may wish to
perform a search of documents 124 and activate search module 112 on
computing device 110. Search module 112 may then activate image
capture module 134 on the device 110 and present the image being
received by image capture module 134 to the user through interface
113. The user may then point the device at a portion of a document
(e.g., an entire document or less than the entire document) and
indicate through interface 113 (e.g., by pressing a button) that
the image being currently presented through the interface 113 is to
be captured and used for a search.
[0059] Search module 112 may then receive the image currently being
displayed through the interface 113 using image capture module 134
and provide the image to feature recognition module 114. Feature
recognition module 114 may then perform OCR on the captured image
to determine text from the image. Search module 112 may then
determine a set of search terms from the determined text. In one
embodiment, to determine a set of search terms from the text
determined from the image, the text may be provided to search term
module 116 which may utilize NLP module 118 to determine one or
more words or phrases from the determined text. Such determinations
may be made, for example, based on frequency, term weighting,
removal of stop words or other NLP techniques.
[0060] Term weighting techniques, for example, include those known
in the art such as term frequency-inverse document frequency
(TF-IDF) involving numerical statistics which reflect an importance
of a term to a particular document in a document group. Search
engines can use TF-IDF as a tool to score and rank term relevance.
TF-IDF may include a value that increases proportionally to the
number of times a word appears in a document, offset by term
frequency in a document group.
[0061] Stop-word processing may be used alone or in conjunction
with TF-IDF to filter out certain terms or phrases prior to NLP.
NLP techniques are further well-known in the art and include, but
are not limited to, automatic summarization, co-reference
resolution, discourse analysis, machine translation, morphological
segmentation, named entity recognition, generation techniques,
part-of-speech tagging, parsing techniques, relationship
extraction, sentence breakdown, sentiment analysis, topic and
speech segmentation, word segmentation, word sense disambiguation,
etc. Once the set of search terms are determined, search module 112
may use interface 113 to provide these search terms to search
interface 128 of search module 126 of content provisioning module
122. For example, the search terms may be provided to the search
module 126 through search interface 128 using an API or web
services interface provided by search interface 128. Search module
126 may then utilize the provided search terms to search the
documents 124 to obtain a set of search results, where the search
results are associated with documents 124 responsive to the
search.
[0062] These results may be returned to interface 113 through
search interface 128 of search module 126 (e.g., in response to the
API or web services call received from interface 113). The results
returned may, for example, be copies of the documents 124 located
through the search (or a subset thereof, such as the top 10 most
responsive or closely matched documents, etc.). Alternatively, an
identifier (e.g., URL, token, document number, etc.) that may be
utilized to locate documents 124 responsive to the search may be
returned as a search result, etc.
[0063] When the results are received by interface 113 they may be
presented to the user at the device 110. For example, a list of the
titles or links to each document returned as a search result (or a
subset thereof) may be presented allowing a user to select the
title and view the corresponding document (e.g., by accessing the
document stored on device 110 or by accessing content provisioning
module 120 with an identifier of the desired document, etc.).
Moreover, in certain embodiments the results of may be displayed
through interface 113 in conjunction with the image of the paper
document on the device that was utilized in the search (e.g., as a
list, set of links, initial sentence of each electronic document,
etc. overlaid on top of the image on the screen at the user's
device 110).
[0064] It should be noted here that while certain embodiments of
presenting the results of a search are discussed and presented
herein, other configurations for presenting the results of a search
may be possible and such configurations are likewise contemplated
by this disclosure.
[0065] For example, in certain embodiments, an interactive search
that may be substantially continuously updated based on the image a
user is currently viewing may be implemented. As an example, in one
embodiment, as a user moves his device over different portions of
the paper document, and a portion of the paper document is
displayed to the user on the user's device, a search is performed
based upon the current portion of the paper document being
displayed on the user's device. The results are displayed to the
user in conjunction with that portion of the document being
currently displayed (e.g., overlaid on top of the image of that
portion of the document on the screen). Thus, as the user moves his
device over different portions of the document the search results
indicating electronic documents associated with those different
portions may be shown to the user and substantially constantly
updated as the user moves his device to those different
portions.
[0066] It can be further noted that, in some embodiments, other
features or aspects of the document, such as formatting, logos,
graphics, a whitespace fingerprint, or the like may be determined
and provided to search module 126 as a search query. For example, a
whitespace fingerprint may be used to allow searching against
patterns associated with the documents in a document group being
searched.
[0067] In such embodiments, a user of computing device 110 may wish
to perform a search of documents 124 and activate search module 112
on computing device 110. Search module 112 may then activate image
capture module 134 on the device 110 and present the image being
received by image capture module 134 to the user through interface
113. Search module 112 may then receive the image currently being
displayed through the interface 113 using image capture module 134,
provide the image to feature recognition module 114, use interface
113 to provide these search terms to search module 126, receive the
search results through interface 113 and present the results
through interface 113.
[0068] The image currently being displayed through the interface
113 may then again be captured using image capture module 134 and
used to search documents 124 and display an updated set of search
results through interface 113. In this manner, as the user moves
the device 110 over a paper document the user may be displayed
search results that are substantially associated with the portion
of the document currently being displayed to the user.
[0069] It should also be noted with respect to embodiments
presented herein, that while in certain embodiments feature
recognition or OCR may be accomplished on mobile computing devices
in conjunction with document searching based on paper document, it
is contemplated by other embodiments that such feature recognition
or OCR may be performed in other locations, such as at platform
120. Thus, for example, a computing device 110 may send search
information to platform 120, where the search information comprises
the image(s) of the document and platform 120 performs the
operations of recognizing features and determining the search
query. In another example, computing device 110 may perform feature
recognition (e.g., OCR) and send the features to platform 120 as
the search information. Platform 120 can determine the search query
from the features provided by client device 110. Thus, the search
information provided by search module 112 may include search terms,
image(s) of a document, features or other information that platform
120 can use to run a search based on the images captured by image
capture module 134. It may be useful here to briefly describe such
embodiments.
[0070] FIG. 2 depicts one embodiment of topology for document
searching where feature recognition is performed on a location
other than the mobile device 210. Here, search module 226 at
platform 220 may include search interface 228, search term module
216 and feature recognition module 214. In this embodiment, search
interface 228 may be configured to receive (e.g., at a particular
URL, through an API or web services interface, etc.) a query
including one or more images. The search module 226 may thus be
configured to use feature recognition module 214 to perform feature
recognition or OCR on the received image to generate text
corresponding to the image and to use search term module 216 to
determine one or more search terms from this text. Search module
226 can then search the documents 224 based on these search terms
to determine a set of results responsive to the query. These
results can then be returned through the search interface 228 in
response to the query.
[0071] Similarly then, in the embodiment depicted, search module
212 may include interface 213, where interface 213 may be
configured to be accessed by a user of the mobile device 210 such
that the image currently being received by image capture module 234
is displayed on the screen of the mobile computer device 210. In
one embodiment, then, interface 213 may allow a user to indicate
that an image (e.g., the image currently being received by the
image capture module) is to be used for a search when activated by
a user (e.g., when the user selects a particular button or touches
a screen in a particular place, etc.). Interface 213 may also be
configured to present the results of a search to a user (e.g., in
conjunction with an image being displayed to the user or in another
type of interface altogether).
[0072] Search module 212 is thus configured to be activated and to
access image capture module 234 to obtain images (e.g., image data)
from image capture module 234 and provide these images to interface
213 to be presented to the user. Search module 212 is also
configured to receive an indication that a user wishes to use an
image currently being displayed for a search. The search module 212
is thus configured to use interface 213 to provide the image to
search module 226 using search interface 228, receive the results
of the search from interface 213 and present the results of the
search using interface 213.
[0073] Accordingly, a user of computing device 210 may wish to
perform a search of documents 224 and activate search module 212 on
computing device 210. Search module 212 may then activate image
capture module 234 on the device 210 and present the image being
received by image capture module 234 to the user through interface
213. The user may then point the device at a portion of a document
(e.g., an entire document or less than the entire document) and
indicate through interface 213 (e.g., by pressing a button) that
the image being currently presented through the interface 213 is to
be captured and used for a search.
[0074] Search module 212 may then receive the image currently being
displayed through the interface 213 using image capture module 234.
Search module 212 may use interface 213 to provide the image to
search interface 228 of search module 226 of content provisioning
module 222. For example, the image may be provided to the search
module 226 through search interface 228 using an API or web
services interface provided by search interface 228.
[0075] The image may be received at search module 226 through
search interface 228. Search module 226 may then provide the image
to feature recognition module 214. Feature recognition module 214
may perform feature recognition such as OCR on the captured image
to determine text or another feature from the image. Search module
226 may receive the text or feature and provide the text or feature
to search term module 216 to determine a set of search terms from
the determined text or feature. In one embodiment, to determine a
set of search terms from the text, the text may be provided to NLP
module 218 to determine one or more words or phrases from the
text.
[0076] Once the set of search terms are determined, search module
226 may utilize the search terms to search the documents 224 to
obtain a set of search results, where the search results are
associated with documents 224 responsive to the search. These
results may be returned to interface 213 through search interface
228 of search module 226 (e.g., in response to the API or web
services call received from interface 213). When the results are
received by interface 213 they may be presented to the user at the
device 210 as discussed above.
[0077] As can be seen then, documents may be searched based on a
paper document in certain embodiments by performing feature
recognition or OCR on a portion of the paper document and using the
resulting text to search the documents, regardless of where the
feature recognition or OCR is performed.
[0078] As noted above, in another non-limiting embodiment, client
computing device can comprise feature recognition module 214. In
this case, search module 212 can send features (OCR text or other
features) to platform 220 and platform 220 can determine search
terms.
[0079] Referring now to FIG. 3, one embodiment of a method for
searching electronic documents based on paper documents using
feature recognition or OCR is depicted. The method of FIG. 3 can be
performed by a client computing device acting in conjunction with a
server (e.g., a server of a platform). Initially, at step 310 an
image is captured on a computing device. This capture may be the
reception of the image (e.g., image data) corresponding to the
image being received through an image capture device such as a
camera or the like on the computing device. The image may be of a
portion of a paper document that is to be used to perform the
search.
[0080] At step 320 feature recognition or OCR may be performed on
the captured image. Performing
[0081] OCR on the captured image may result in text associated with
the image of the portion of the paper document (e.g., text that was
included in the captured image). In other embodiments, document
layout or document formatting may be identified. Using the results
of the feature recognition, such as text resulting from the OCR of
the captured image, a set of search terms may be determined at step
330. By way of additional example, a document layout or document
formatting may be determined and used as search terms to search for
comparable documents.
[0082] In other embodiments, such as those using OCR, search terms
may, for example, be determined using NLP techniques or the like to
remove stop words from the text, determine key words or phrases,
performing key word or term weighting, etc. Thus, the set of search
terms may be determined from the text by, for example, using all
the text (e.g., each term of the text is a search term, or the
entire text is one search term), using a set of key words or
phrases determined form the text, using the words of the text after
removal of stop words or by determining the set of search terms in
some other manner.
[0083] A set of electronic documents can then be searched based on
the set of search terms at step 340.
[0084] It will be realized that the set of documents may be
searched based on an index of those documents or the documents
themselves may be searched or some other method may be used to
search the documents. Thus, embodiments as presented herein may be
utilized with equal efficacy in cases where documents are resident
in one or more databases such as a database system or content
management system or in cases where documents are distributed
across a network and an index of such documents is maintained such
as in an Internet or internet search engine, etc.
[0085] Once the search is performed, the results of that search may
be presented at step 350. For example, a list of the titles or
links to each document returned as a search result (or a subset
thereof) may be presented through a display of the computing device
in a manner that allows a user to select the title and view the
corresponding document. Moreover, in certain embodiments the
results of the search may be displayed through an interface in
conjunction with the image of the paper document on the device that
was utilized in the search (e.g., as a list, set of links, etc.).
In one embodiment, another image may then again be captured at step
310 and used to search documents such that the method is repeated
and the search results updated. The steps may be repeated and the
search results updated until, for example, the user selects a
particular document of the search results to view or otherwise
indicates an end to the search.
[0086] As discussed above, many of the images used to perform
embodiments as presented herein may be captured and processed on a
mobile device. However, currently a number of impediments to doing
feature recognition or OCR on mobile devices exist, including, for
example, relatively inferior camera performance or sensors, the
ability of current feature recognition or OCR technologies to deal
with images that are skewed or rotated, the large processing power
required to perform feature recognition or OCR on a captured image
etc. Thus, in many cases performing feature recognition or OCR of a
document on a mobile device may take an inordinately long time or
may fail altogether.
[0087] Despite these impediments, however, it is still desired to
perform image capture and feature recognition or OCR on mobile
devices. Accordingly, systems and methods to effectively implement
feature recognition or OCR on a mobile device are presented herein.
Specifically, according to certain embodiments, a set of images of
a document may be captured, where each image may be of a portion of
the document, where the portion is less than the entire document.
Feature recognition or OCR may be performed on each of these images
and the text resulting from performing feature recognition or OCR
on each of the images joined to form an overall text corresponding
to the images of those portions of document. Thus, a single overall
text for a document may be generated more efficiently by performing
feature recognition or OCR on images of a set of portions of the
document and joining them together.
[0088] In one embodiment, when the text corresponding to each image
of each portion is obtained by OCR the resulting text may be joined
by performing language processing (e.g., natural language
processing) techniques to analyse the texts obtained from the
images to determine how they are to be joined. For example,
overlapping text in different images may be used to determine which
text should be joined and where.
[0089] In some non-limiting embodiments, joining and overlapping
text can be determined by searching for commonly found terms/words
in different portions of OCR text. Commonly found terms can include
matching words, phrases, numbers, characters, symbols and/or tokens
such as the sentence tokens such as commas, periods, etc. In some
more specific embodiments, a starting text can be identified in a
first document and/or OCR portion such that the starting text may
be searched for in a second document portion and/or OCR portion by
searching each identified text from the second document portion
(i.e., a word-by-word comparison search). Upon discovery of a
match, a further match can be extended by searching for matching
surrounding words. The number of iterations in a surrounding word
search may be predetermined. Different portions of texts may be
searched as such until a match is found sufficient to join the
texts or until all possible text portion combinations are
compared.
[0090] Additionally, in some embodiment, in a similar manner to how
predictive text on mobiles works, with an appropriate language
database on the device, text from the images could be joined by
evaluating or determining a probability of texts from different
images matching.
[0091] Additionally, in certain embodiments, other data that may be
obtained from the mobile device may be used to inform the joining
process. For example, motion data associated with the set of images
such as the accelerometer or other motion sensor on the mobile
device may be used to determine a direction of movement of the
mobile device when the set of images were captured or orientation
of the device when the set of images were captured, etc. This
motion data may be used when joining the text obtained from the set
of images and may further improve both the results and the speed of
the joining process. For example, the motion data associated with
the set of images may indicate an order of the images. Documents
can be identified for further processing based on relative
movements of pixels on the screen, in some cases a group of pixels
identified as corresponding to a distinct document. Relative
movements may take into account angular dispositions (i.e., from
different oblique views), relative dimensional movements, etc.
[0092] Alternatively, or additionally, in certain embodiments
motion data such as movement between images or direction of travel
of the device over the document) may be determined algorithmically
(e.g., mathematically) from the image data and the determined
motion data used during the joining process. In some instances, an
accelerometer may be used to determine device travel as well as
device rotation in order to map such information to a different set
of captured images so that such data can be used to analyse and
form a basis for document recognition.
[0093] It may now be helpful here to discuss embodiments of various
topologies that may be utilized in performing embodiments as
disclosed herein. Referring now to FIG. 4, one embodiment of a
mobile device 410 configured to perform OCR is depicted. Mobile
device 410 includes image capture module 434, motion detection
module 436 and joining module 412.
[0094] Image capture module 434 may access a lens or other aperture
configured to receive or capture images using the light received
through the aperture, as discussed above. Motion detection module
436 may be a device configured to detect motion, speed, direction,
orientation or other data associated with position or movement of
mobile device 410, with the assistance of an accelerometer or the
like.
[0095] Joining module 412 may include OCR module 414, language
processing module 416 and telemetry module 418. OCR module 414 is
configured to perform OCR on an image, where the results of the OCR
of an image is text corresponding to that image.
[0096] Language processing module 416 may be configured to perform
language processing (e.g., natural language processing) techniques
to analyse how, and if, text is to be joined. For example,
overlapping text (e.g., words, phrases, portions of words or
phrases, numbers, etc.) in the text obtained from different images
may be used to determine which texts should be joined and where
these texts should be joined. For example, the language processing
module 416 may be configured to perform the search and evaluation
techniques as described above. Additionally, in some embodiment, in
a similar manner to how predictive text on mobiles works (e.g.,
using an appropriate language database) language processing module
416 may determine a probability of a match between two texts by
analysing the likelihood of one text (or portion thereof) following
another text (or portion thereof).
[0097] Telemetry module 418 may be configured to receive or
determine motion, speed, direction, orientation, etc. data
(collectively motion data) associated with images received at
joining module. Telemetry module 418 may utilize motion data
received directly from motion detection module 436 in conjunction
with an image or determine such motion data (e.g., algorithmically)
from, for example, image data or other motion data associated with
the images. The telemetry module 418 may be configured to
determine, based on the motion data received or obtained, an order
of the images received at joining module 412.
[0098] Joining module 412 may thus be configured to receive a set
of images 442 associated with a document 440 from image capture
module 434 along with, in some embodiments, motion data associated
with the images from motion detection module 436. Joining module
412 may be further configured to provide such images to OCR module
414 to obtain text associated with each of the images. Joining
module 412 may then be configured to use language processing module
416 to determine which of the texts to join (if any) and where to
join the text. Joining module 412 may also, in some embodiments,
use motion data from telemetry module 416 corresponding to the
images from which the texts were obtained to determine which of the
text to join or where to joining the texts.
[0099] Once joining module 412 has determined which texts are to be
joined and where those texts are to be joined, joining module 412
may join the texts to produce a single overall text corresponding
to images 442 from the document 440. When joining the texts (e.g.,
at least two of the texts), joining module 412 may not utilize all
the texts, may utilize only portions of certain texts, may remove
duplicative (e.g., overlapping) text, etc.
[0100] One embodiment of a method for performing OCR on devices
using multiple images is depicted in FIG. 5. Initially at step 510
a set of images may be captured form a document. This capture may
be the reception of images (e.g., image data) corresponding to the
images being received through an image capture device such as a
camera or the like on the computing device. The images received may
be captured based on input from the user (e.g., the user may point
the computing device and select when an image is to be captured) or
may be captured at certain intervals as, for example, a user moves
the device over the document.
[0101] At step 520 OCR may be performed on the captured images.
Performing OCR on the captured image may result in a text
associated with each captured image. At step 530 a determination
can be made which of those texts are to be joined together. This
determination can be made, in one embodiment, by analyzing each of
the texts to determine portions of the texts (e.g., words, phrases,
numbers, etc.) that overlap (e.g., are the same or are a part of
the same word, phrase or number, etc.). This analysis may also take
into account the overlap between texts that occur at different
lines of the text (e.g., an overlap between two texts may occur at
multiple places and on multiple lines of the texts).
[0102] In some embodiments, the determination of which texts are to
be joined may be made based on motion data. For example if it can
be determined that a device was moving from left to right when
image 1 and image 2 were taken, it may be the case that the text
resulting from the OCR of image 1 should be joined to the text
resulting from the OCR of image 2.
[0103] Additionally, predictive linguistics may be used to
determine which texts should be joined. Such predictive linguistics
may be used, for example, to determine a likelihood that a word or
phrase of one text would follow a word of phrase of another
text.
[0104] Once a determination is made which texts should be joined,
these texts may be joined at step 540. This joining process may
join texts in such a way that duplicative text may be removed and
at least two of the texts (or portions thereof) joined to form a
single overall text corresponding to the captured images.
[0105] The joining of texts obtained from different images taken
from the same document may be better understood with reference to
FIG. 6. In the example illustrated, images 610 may be captured from
document 600. Areas 620 may include overlapping text that may be
obtained from each of these images 610 that may be used to
determine which texts to join. For example, notice that image 610a
incudes the text "t commune. Te quo nominavi" and "haedrum perpetua
id. Vel" while image 610c includes the text "d est, duo te feugiat
commun" and "peratoribus, mea phaedrum p". Thus, as depicted in
area 620d the text of image 610c and the text of image 610a may be
determined to comprise the overlapping text "t commun" and "haedrum
p". Accordingly, it can be determined that the text from these
images 610a, 610c should be joined based on this overlapping text
and when the text of the images 610a, 610c are joined it may result
in the text "d est, duo te feugiat commune. Te quo nominavi" and
"peratoribus, mea phaedrum perpetua id. Vel".
[0106] Similarly, then, an OCR of image 610d includes the text
"Pericula gloriatur ad est," and "mediocritatem vituperato" while
image 610c includes the text "d est, duo te feugiat commun" and
"peratoribus, mea phaedrum p". Thus, as depicted in area 620b the
text of image 610d and the text of image 610c may be determined to
comprise the overlapping text "d est," and "perato". Accordingly,
it can be determined that the text from these images 610d, 610c
should be joined based on this overlapping text and, when the text
of the images 610d, 610c are joined it may result in the text
"Pericula gloriatur ad est, duo te feugiat commun" and
"mediocritatem vituperatoribus, mea phaedrum p". Furthermore, as
discussed above, it may have been determined that the text from
images 610c and 610a should also be joined based on overlapping
text within those images. As such, when the text from image 610d is
joined to the text from image 610c and the text from image 610c is
joined to the text from image 610a it may result in the text
"Pericula gloriatur ad est, duo te feugiat commune. Te quo
nominavi" and "mediocritatem vituperatoribus, mea phaedrum perpetua
id. Vel".
[0107] Examples of text that may be used to join text of image 610d
to text of image 610b is depicted in area 620a and examples of text
that may be used to join text of image 610b to text of image 610a
is depicted in area 620c.
[0108] Referring to FIG. 7, another embodiment of topology for OCR
of documents is illustrated. The topology includes one or more
computing devices 710 connected to a content provisioning platform
720 over a network. The network may be a wired or wireless network
such as the Internet, an intranet, a LAN, a WAN, a cellular
network, another type of network. It will be understood that the
network may be a combination of multiple different kinds of wired
or wireless networks. In this embodiment, however, the joining is
performed at platform 720. Mobile device 710 includes image capture
module 734, OCR module 714, and motion detection module 736.
[0109] Image capture module 734 may access a lens or other aperture
configured to receive or capture images using the light received
through the aperture, as discussed above. Motion detection module
736 may be a device configured to detect motion, speed, direction,
orientation or other data associated with position or movement of
mobile device 710, with the assistance of an accelerometer or the
like. OCR module 714 is configured to perform OCR on an image,
where the results of the OCR of an image is text corresponding to
that image.
[0110] The back end platform 720 may receive the image capture,
OCR, and motion detection data from the mobile device 710 over a
network, such as the Internet or the like. The back end platform
720 includes a joining module 712. Joining module 712 may include
language processing module 716 and telemetry module 718.
[0111] Language processing module 716 may be configured to perform
language processing (e.g., natural language processing) techniques
to analyze how, and if, text is to be joined. For example,
overlapping text (e.g., words, phrases, portions of words or
phrases, numbers, etc.) in the text obtained from different images
may be used to determine which texts should be joined and where
these texts should be joined. For example, the language processing
module 716 may be configured to perform the search and evaluation
techniques as described above. Additionally, in some embodiment, in
a similar manner to how predictive text on mobiles works (e.g.,
using an appropriate language database) language processing module
716 may determine a probability of a match between two texts by
analyzing the likelihood of one text (or portion thereof) following
another text (or portion thereof).
[0112] Telemetry module 718 may be configured to receive or
determine motion, speed, direction, orientation, etc. data
(collectively motion data) associated with images received at the
joining module. Telemetry module 718 may utilize motion data
received directly from motion detection module 736 in conjunction
with an image or determine such motion data (e.g., algorithmically)
from, for example, image data or other motion data associated with
the images. The telemetry module 718 may be configured to
determine, based on the motion data received or obtained, an order
of the images received at joining module 712.
[0113] Joining module 712 may thus be configured to receive a set
of images 742 associated with a document 740 from image capture
module 734 along with, in some embodiments, motion data associated
with the images from motion detection module 736, as well as text
associated with each of the images from the OCR module 714. Joining
module 712 may then be configured to use language processing module
716 to determine which of the texts to join (if any) and where to
join the text. Joining module 712 may also, in some embodiments,
use motion data from telemetry module 716 corresponding to the
images from which the texts were obtained to determine which of the
text to join or where to joining the texts.
[0114] Once joining module 712 has determined which texts are to be
joined and where those texts are to be joined, joining module 712
may join the texts to produce a single overall text corresponding
to images 742 from the document 740. When joining the texts (e.g.,
at least two of the texts), joining module 712 may not utilize all
the texts, may utilize only portions of certain texts, may remove
duplicative (e.g., overlapping) text, etc.
[0115] Another embodiment of a topology for OCR of documents is
illustrated in FIG. 8. The topology includes one or more computing
devices 810 connected to a content provisioning platform 820 over a
network. The network may be a wired or wireless network such as the
Internet, an intranet, a LAN, a WAN, a cellular network, another
type of network. It will be understood that the network may be a
combination of multiple different kinds of wired or wireless
networks. In this embodiment, the OCR and joining are performed at
platform 820. Mobile device 810 includes image capture module 834,
OCR module 814, and motion detection module 836.
[0116] Mobile device 810 includes image capture module 834 and
motion detection module 836. Image capture module 834 may access a
lens or other aperture configured to receive or capture images
using the light received through the aperture, as discussed above.
Motion detection module 836 may be a device configured to detect
motion, speed, direction, orientation or other data associated with
position or movement of mobile device 810, with the assistance of
an accelerometer or the like.
[0117] The back end platform 820 may receive the image capture and
motion detection data from the mobile device 810 over a network,
such as the Internet or the like. The back end platform 820
includes a joining module 812. Joining module 812 may include
language processing module 816, OCR module 814, and telemetry
module 818. OCR module 814 is configured to perform OCR on an
image, where the result of the OCR of an image is text
corresponding to that image.
[0118] Joining module 812 may include OCR module 814, language
processing module 816 and telemetry module 818. OCR module 814 is
configured to perform OCR on an image, where the results of the OCR
of an image is text corresponding to that image.
[0119] Language processing module 816 may be configured to perform
language processing (e.g., natural language processing) techniques
to analyze how, and if, text is to be joined. For example,
overlapping text (e.g., words, phrases, portions of words or
phrases, numbers, etc.) in the text obtained from different images
may be used to determine which texts should be joined and where
these texts should be joined. For example, the language processing
module 816 may be configured to perform the search and evaluation
techniques as described above. Additionally, in some embodiment, in
a similar manner to how predictive text on mobiles works (e.g.,
using an appropriate language database) language processing module
816 may determine a probability of a match between two texts by
analyzing the likelihood of one text (or portion thereof) following
another text (or portion thereof).
[0120] Telemetry module 818 may be configured to receive or
determine motion, speed, direction, orientation, etc. data
(collectively motion data) associated with images received at
joining module. Telemetry module 818 may utilize motion data
received directly from motion detection module 836 in conjunction
with an image or determine such motion data (e.g., algorithmically)
from, for example, image data or other motion data associated with
the images. The telemetry module 818 may be configured to
determine, based on the motion data received or obtained, an order
of the images received at joining module 812.
[0121] Joining module 812 may thus be configured to receive a set
of images 842 associated with a document 840 from image capture
module 834 along with, in some embodiments, motion data associated
with the images from motion detection module 836. Joining module
812 may be further configured to provide such images to OCR module
814 to obtain text associated with each of the images. Joining
module 812 may then be configured to use language processing module
816 to determine which of the texts to join (if any) and where to
join the text. Joining module 812 may also, in some embodiments,
use motion data from telemetry module 816 corresponding to the
images from which the texts were obtained to determine which of the
text to join or where to joining the texts.
[0122] Once joining module 812 has determined which texts are to be
joined and where those texts are to be joined, joining module 812
may join the texts to produce a single overall text corresponding
to images 842 from the document 840. When joining the texts (e.g.,
at least two of the texts), joining module 812 may not utilize all
the texts, may utilize only portions of certain texts, may remove
duplicative (e.g., overlapping) text, etc.
[0123] It is noted that various other configurations of the joining
and OCR may be implemented. For example, in some embodiments, the
image and/or telemetry data may be sent from the mobile device 810
to the platform 420, which then performs the OCR. The platform may
then send the results of the OCR and other analysis to the user
device 810 which may then perform the joining.
[0124] As can be seen then, certain embodiments may provide
improved methods for performing OCR on computing devices. Such
embodiments may be usefully applied in systems and methods for
document searching. FIG. 9 depicts one embodiment of a topology for
document searching that includes the joining of texts. The topology
includes one or more computing devices 910 connected to a content
provisioning platform 920 over a network 930. The network 930 may
be a wired or wireless network such as the Internet, an intranet, a
LAN, a WAN, a cellular network, another type of network. It will be
understood that network 930 may be a combination of multiple
different kinds of wired or wireless networks.
[0125] Platform 920 may include one or more servers providing
content provisioning modules 922 accessible at one or more
locations (e.g., IP addresses or domain names) or through one or
more interfaces. The content provisioning module 922 of a platform
920 may support access from a computing device 910. Content
provisioning module 922 may include search module 926 including
search interface 928 to allow for searching of such documents 924.
Search interface 928 may be accessible (e.g., at a particular URL,
through an API or web services interface, etc.) such that a query
including one or more search terms may be provided through the
search interface 928. The search module 926 may search the
documents 924 based on the query provided through the search
interface 926 to determine a set of results responsive to the
query. These results can then be returned by the search interface
928 in response to the query.
[0126] Each computing device may include an image capture module
934, motion detection module 936 and a search module 912. Image
capture module 934 may include a lens or other aperture and be
configured to receive or capture images using the light received
through the aperture. For example, many mobile computing devices
include a cameras lens or the like and an associated image capture
module. Thus, image capture module 934 may, when utilized, receive
image data (with or without direct user involvement) through the
aperture of the device 910 and capture or otherwise provide these
images. Motion detection module 936 may be a device configured to
detect motion, speed, direction, orientation or other data
associated with position or movement of mobile device 910, such as
an accelerometer or the like.
[0127] Search module 912 may include interface 913, joining module
950 and search term module 917. Interface 913 may have access to
one or more input or output modules (e.g., a screen, buttons,
speakers, etc.) on device 910 and be configured to provide output
or receive input through these input or output modules of the
device 910. Interface 913 may also be configured to utilize other
interfaces such as APIs, web service interfaces, etc. to send
requests and receive responses to those requests.
[0128] Specifically, in one embodiment, interface 913 may be
configured to be accessed by a user such that the image currently
being received by image capture module 934 is displayed on the
screen of the mobile computer device 910. In one embodiment, then,
interface 913 may allow a user to indicate that images (e.g.,
images captured while a user is moving the device 910 over the
document) is to be used for a search when activated by a user
(e.g., when the user selects a particular button or touches a
screen in a particular place, etc.). Interface 913 may also be
configured to present the results of a search to a user (e.g., in
conjunction with an image being displayed to the user or in another
type of interface altogether).
[0129] Search term module 917 is configured to determine one or
more search terms from text. Search term module 917 may utilize,
for example, language processing module 916 or the like configured
to remove stop words from the text, determine key words or phrases,
performing key word or term weighting, etc. Thus, search term
module 917 may determine a set of search terms from text by, for
example, using all the text (e.g., each term of the text is a
search term, or the entire text is one search term), using a set of
key words or phrases determined form the text, using the words of
the text after removal of stop words, or by determining the set of
search terms in some other manner.
[0130] Joining module 912 may include OCR module 914, language
processing module 916 and telemetry module 918. OCR module 914 is
configured to perform OCR on an image, where the results of the OCR
of an image is text corresponding to that image.
[0131] Language processing module 916 may be configured to perform
language processing (e.g., natural language processing) techniques
to analyse how, and if, text is to be joined. For example,
overlapping text (e.g., words, phrases, portions of words or
phrases, numbers, etc.) in the text obtained from different images
may be used to determine which texts should be joined and where
these texts should be joined. Additionally, in some embodiment, in
a similar manner to how predictive text on mobiles works (e.g.,
using an appropriate language database) language processing module
916 may determine a probability of a match between two texts by
analysing the likelihood of one text following another text.
[0132] Telemetry module 918 may be configured to receive or
determine motion, speed, direction, orientation, etc. data
(collectively motion data) associated with images received at
joining module. Telemetry module 918 may utilize motion data
received directly from motion detection module 936 in conjunction
with an image or determine such motion data (e.g., algorithmically)
from, for example, image data or other motion data associated with
the images.
[0133] Joining module 912 may thus be configured to receive a set
of images associated with a document along with, in some
embodiments, motion data associated with the images from motion
detection module 936. Joining module 912 may be further configured
to provide such images to OCR module 914 to obtain text associated
with each of the images. Joining module 912 may then be configured
to use language processing module 916 to determine which of the
texts to join (if any) and where to join the text. Joining module
912 may also, in some embodiments, use motion data from telemetry
module 918 corresponding to the images from which the texts were
obtained to determine which of the text to join or where to joining
the texts.
[0134] Search module 912 is thus configured to be activated and to
access image capture module 934 to obtain images (e.g., image data)
from image capture module 934 and provide these images to interface
913 to be presented to the user. Search module 912 is also
configured to receive an indication that a user wishes to perform a
search based on the document (e.g., as the user moves his device
910 over the document). The search module 912 is thus configured to
provide multiple images to joining module 950, receive the overall
text associated with the multiple images from the joining module
950, provide the text to search term module 917, receive the search
terms from search term module 917, use interface 913 to provide the
search terms to search module 926, receive the results of the
search from interface 913 and present the results of the search
using interface 913.
[0135] Accordingly, a user of computing device 910 may wish to
perform a search of documents 924 and activate search module 912 on
computing device 910. Search module 912 may then activate image
capture module 934 on the device 910 and present the image begin
received by image capture module 934 to the user through interface
913. The user may then move the device 910 over the document. At a
certain interval then, the search module 112 may capture multiple
images received from image capture module 934.
[0136] Search module 912 may then provide these multiple images to
joining module 950 which may then OCR on each of these images top
determine text for each of these images and determine if and where
these text should be joined and join the texts accordingly to
generate a single overall text. The single overall text may be
provided to search module 912. Search module 912 may then determine
a set of search terms from the single overall text.
[0137] Once the set of search terms are determined, search module
912 may use interface 913 to provide these search terms to search
interface 928 of search module 926 of content provisioning module
922. For example, the search terms may be provided to the search
module 926 through search interface 928 using an API or web
services interface provided by search interface 928. Search module
926 may then utilize the provided search terms to search the
documents 924 to obtain a set of search results, where the search
results are associated with documents 924 responsive to the search.
These results may be returned to interface 913 through search
interface 928 of search module 926 (e.g., in response to the API or
web services call received from interface 913).
[0138] When the results are received by interface 913 they may be
presented to the user at the device 910. For example, a list of the
titles or links to each document returned as a search result (or a
subset thereof) may be presented allowing a user to select the
title and view the corresponding document (e.g., by accessing the
document stored on device 910 or by accessing content provisioning
module 920 with an identifier of the desired document, etc.).
[0139] It can be appreciated that other search systems may also be
implemented. In some embodiments, the platform can provide one or
more of a joining module, search module, search term module,
feature recognition module, or NLP module. The client computing
device sending search information to the platform, where the search
information includes information to be processed into a search
(e.g., one or more of an image, feature of a document, search term
or other information).
[0140] Routines, methods, functions, steps, operations or portions
thereof described herein can be implemented through control logic
adapted to direct a computing device to perform the routines,
methods, functions, steps, operations or portions thereof. Control
logic can include computer executable instructions stored on a
computer readable medium that can be operated on by a processor,
hardware, firmware or a combination thereof. The control logic can
include, in some embodiments, application specific integrated
circuits, programmable logic devices, field programmable gate
arrays, optical, chemical, biological, quantum or nanoengineered
systems, components and mechanisms. Any suitable language can be
used. Different programming techniques can be employed such as
procedural or object oriented. Based on the disclosure and
teachings provided herein, a person of ordinary skill in the art
will appreciate other ways and/or methods to implement the
invention.
[0141] Any particular step, operation, method, routine, operation
or portion thereof can execute on a single computer processing
device or multiple computer processing devices, a single computer
processor or multiple computer processors. Data may be stored in a
single storage medium or distributed through multiple storage
mediums, and may reside in a single database or multiple databases
(or other data storage). The sequence of operations described
herein can be interrupted, suspended, or otherwise controlled by
another process, such as an operating system, kernel, etc.
[0142] A "computer-readable medium" may be any type of data storage
medium that can store computer instructions, including, but not
limited to read-only memory (ROM), random access memory (RAM), hard
disks (HD), data cartridges, data backup magnetic tapes, floppy
diskettes, flash memory, optical data storage, CD-ROMs, or the
like. The computer readable medium may include multiple computer
readable media storing computer executable instructions, such as in
a distributed system or instructions stored across an array.
[0143] A "processor" includes any hardware system, hardware
mechanism or hardware component that processes data, signals or
other information. A processor can include a system with a central
processing unit, multiple processing units, dedicated circuitry for
achieving functionality, or other systems. A processor can perform
its functions in "real-time," "offline," in a "batch mode," etc.
Portions of processing can be performed at different times and at
different locations, by different (or the same) processing
systems.
[0144] It will be understood for purposes of this disclosure that a
service or module is one or more computer devices, configured
(e.g., by a computer process or hardware) to perform one or more
functions. A service may present one or more interfaces which can
be utilized to access these functions. Such interfaces include
APIs, interfaces presented for a web services, web pages, remote
procedure calls, remote method invocation, etc.
[0145] Communications between computers implementing embodiments of
the invention can be accomplished using any electronic, optical,
radio frequency signals, or other suitable methods and tools of
communication in compliance with network and other communications
protocols.
[0146] As used herein, the terms "comprises," "comprising,"
"includes," "including," "has," "having" or any other variation
thereof, are intended to cover a non-exclusive inclusion. For
example, a process, article, or apparatus that comprises a list of
elements is not necessarily limited to only those elements but may
include other elements not expressly listed or inherent to such
process, article, or apparatus.
[0147] Further, unless expressly stated to the contrary, "or"
refers to an inclusive or and not to an exclusive or. That is, the
term "or" as used herein is generally intended to mean "and/or"
unless otherwise indicated. For example, a condition A or B is
satisfied by any one of the following: A is true (or present) and B
is false (or not present), A is false (or not present) and B is
true (or present), and both A and B are true (or present).
[0148] As used herein, a term preceded by "a" or "an" (and "the"
when antecedent basis is "a" or "an") includes both singular and
plural of such term unless the context clearly dictates otherwise.
Also, as used in the description herein, the meaning of "in"
includes "in" and "on" unless the context clearly dictates
otherwise.
[0149] Additionally, any examples or illustrations given herein are
not to be regarded in any way as restrictions on, limits to, or
express definitions of, any term or terms with which they are
utilized. Instead, these examples or illustrations are to be
regarded as being described with respect to one particular
embodiment and as illustrative only. Those of ordinary skill in the
art will appreciate that any term or terms with which these
examples or illustrations are utilized will encompass other
embodiments which may or may not be given therewith or elsewhere in
the specification and all such embodiments are intended to be
included within the scope of that term or terms. Language
designating such nonlimiting examples and illustrations includes,
but is not limited to: "for example," "for instance," "e.g.," "in
one embodiment."
[0150] Reference throughout this specification to "one embodiment,"
"an embodiment," or "a specific embodiment" or similar terminology
means that a particular feature, structure, or characteristic
described in connection with the embodiment is included in at least
one embodiment and may not necessarily be present in all
embodiments. Thus, respective appearances of the phrases "in one
embodiment," "in an embodiment," or "in a specific embodiment" or
similar terminology in various places throughout this specification
are not necessarily referring to the same embodiment. Furthermore,
the particular features, structures, or characteristics of any
particular embodiment may be combined in any suitable manner with
one or more other embodiments. Moreover, it will be appreciated
that in some instances some features of embodiments of the
invention will be employed without a corresponding use of other
features without departing from the scope and spirit of the
invention as set forth.
[0151] In the description herein, numerous specific details are
provided, such as examples of components and/or methods, to provide
a thorough understanding of embodiments of the invention. One
skilled in the relevant art will recognize, however, that an
embodiment may be able to be practiced without one or more of the
specific details, or with other apparatus, systems, assemblies,
methods, components, materials, parts, and/or the like. In other
instances, well-known structures, components, systems, materials,
or operations are not specifically shown or described in detail to
avoid obscuring aspects of embodiments of the invention. While the
invention may be illustrated by using a particular embodiment, this
is not and does not limit the invention to any particular
embodiment and a person of ordinary skill in the art will recognize
that additional embodiments are readily understandable and are a
part of this invention.
[0152] Although the steps, operations, or computations may be
presented in a specific order, this order may be changed in
different embodiments. In some embodiments, to the extent multiple
steps are shown as sequential in this specification, some
combination of such steps in alternative embodiments may be
performed at the same time. The sequence of operations described
herein can be interrupted, suspended, or otherwise controlled by
another process.
[0153] It will also be appreciated that one or more of the elements
depicted in the drawings/figures can also be implemented in a more
separated or integrated manner, or even removed or rendered as
inoperable in certain cases, as is useful in accordance with a
particular application. Additionally, any signal arrows in the
drawings/figures should be considered only as exemplary, and not
limiting, unless otherwise specifically noted.
[0154] Benefits, other advantages, and solutions to problems have
been described above with regard to specific embodiments. However,
the benefits, advantages, solutions to problems, and any
component(s) that may cause any benefit, advantage, or solution to
occur or become more pronounced are not to be construed as a
critical, required, or essential feature or component.
[0155] It is to be understood that other variations and
modifications of the embodiments described and illustrated herein
are possible in light of the teachings herein and are to be
considered as part of the spirit and scope of the invention. Thus,
while the invention has been described herein with reference to
particular embodiments thereof, a latitude of modification, various
changes and substitutions are intended in the foregoing
disclosures, and therefore, many modifications may be made to adapt
a particular situation or material to the essential scope and
spirit of the invention. Accordingly, the specification, including
the Summary and Abstract, and figures are to be regarded in an
illustrative rather than a restrictive sense, and all such
modifications are intended to be included within the scope of
invention.
* * * * *