U.S. patent application number 11/710395 was filed with the patent office on 2007-07-26 for texture-based image database browsing and sorting method.
This patent application is currently assigned to Electronics and Telecommunications Research Institute. Invention is credited to Young Lae Bae, Byung Tea Chun, Se Yoon Jeong, Kyu Heon Kim, Jae Yeon Lee.
Application Number | 20070174280 11/710395 |
Document ID | / |
Family ID | 19706937 |
Filed Date | 2007-07-26 |
United States Patent
Application |
20070174280 |
Kind Code |
A1 |
Kim; Kyu Heon ; et
al. |
July 26, 2007 |
Texture-based image database browsing and sorting method
Abstract
A texture-based image database browsing and sorting method
computes the number of edge pixels of objects in static images,
measures textures of the static images by numerating the number of
edge pixels thereof and measures a texture of a query image by
numerating the number of edge pixels of an object in the query
image. Then, the method sorts the measured textures according to a
sorting order and searches a texture close to the texture of the
query image among the sorted textures.
Inventors: |
Kim; Kyu Heon; (Taejon,
KR) ; Jeong; Se Yoon; (Taejon, KR) ; Lee; Jae
Yeon; (Taejon, KR) ; Bae; Young Lae; (Taejon,
KR) ; Chun; Byung Tea; (Taejon, KR) |
Correspondence
Address: |
LOWE HAUPTMAN BERNER, LLP
1700 DIAGONAL ROAD
SUITE 300
ALEXANDRIA
VA
22314
US
|
Assignee: |
Electronics and Telecommunications
Research Institute
|
Family ID: |
19706937 |
Appl. No.: |
11/710395 |
Filed: |
February 26, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
09995738 |
Nov 29, 2001 |
7188099 |
|
|
11710395 |
Feb 26, 2007 |
|
|
|
Current U.S.
Class: |
1/1 ;
707/999.007; 707/E17.025 |
Current CPC
Class: |
Y10S 707/99937 20130101;
Y10S 707/915 20130101; Y10S 707/99943 20130101; G06F 16/5862
20190101; Y10S 707/99942 20130101; Y10S 707/99933 20130101 |
Class at
Publication: |
707/007 |
International
Class: |
G06F 7/00 20060101
G06F007/00 |
Foreign Application Data
Date |
Code |
Application Number |
Mar 14, 2001 |
KR |
2001-13209 |
Claims
1. A method for sorting static images, comprising the steps of:
counting a number of edge pixels of objects in each of the static
images to measure a texture of each of the static images, the
measured texture of each of the static images being defined as the
counted number of edge pixels of objects in each of the static
images; and sorting the static images based on the measured
textures of each of the static images.
2. The method of claim 1, wherein the step of measuring includes
the steps of: performing an n-dimensional Wavelet transform by
using a high pass filter so as to obtain transformed static images;
and performing a Soble operator in order to eliminate noises
included in the transformed static images and counting the number
of edge pixels of the objects in the transformed static images.
3. The method of claim 1, wherein the step of sorting sorts the
measured textures such that values thereof are in an order of
closeness to a texture value of a newly inputted texture.
4-5. (canceled)
6. A method for sorting static images, comprising the steps, in the
following order, of: a) performing an n-dimensional Wavelet
transfor for transforming the static images by using a high pass
filter; b) performing a Sobel operator for eliminating noise
included in the transformed static images; c) counting edge pixels
of objects in each of the static images; d) representing a texture
of each of the static images with a numeric value representing the
counted number of edge pixels; and e) sorting the static images
according to the numeric values of the static images; respectively.
Description
CROSS REFERENCE TO RELATED APPLICATION
[0001] This application is a divisional of U.S. Ser. No.
09/995,738, filed on Nov. 29, 2001. This application, in its
entirety, is incorporated herein by reference.
[0002] 1. Field of the Invention
[0003] The present invention relates to a method for browsing and
sorting image database based on texture of images; and, more
particularly, to a texture-based image database browsing and
sorting method capable of effectively sorting and browsing a large
amount of image data by using textures of static images, wherein a
texture is obtained by determining the number of edge pixels of an
object in the static image.
[0004] 2. Background of the Invention
[0005] Since existing image browsing or sorting techniques browse
or sort images by using apparent characteristics of a target image,
e.g., a size of image data or a generation time of the image, it is
impossible to practically perform intellectual searching or sorting
by using meaningful information obtained from internal information
included in each image, e.g., color or nature of the image, or an
object in the image.
SUMMARY OF THE INVENTION
[0006] It is, therefore, a primary object of the present invention
to provide a texture-based image database browsing and sorting
method for effectively browsing or sorting images without examining
individual image data by measuring texture which is obtained by
computing the number of edge pixels of an object within a target
static image through the use of a Sobel operator and an
n-dimensional Wavelet transform using a high frequency filter.
[0007] In accordance with a preferred embodiment of the present
invention, there is provided a method for sorting static images,
comprising the steps of:
[0008] computing the number of edge pixels of objects in the static
images and measuring textures of the static image by numerating the
number of edge pixels; and
[0009] sorting the measured textures according to a sorting
order.
[0010] In accordance with another preferred embodiment of the
present invention, there is provided a method for browsing static
images in a data image texture database by using an inputted query
image as a standard, comprising the steps of:
[0011] measuring a texture of the query image by numerating the
number of edge pixels of an object in the query image; and
[0012] searching a texture value close to the texture of the query
image among the textures of the static images sorted in the data
image texture database.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] The above and other objects and features of the present
invention will become apparent from the following description of
preferred embodiments given in conjunction with the accompanying
drawings, in which:
[0014] FIG. 1 shows a schematic diagram of explaining an image
database sorting method in accordance with the present
invention;
[0015] FIG. 2 illustrates a schematic diagram of showing an image
database browsing method in accordance with the present
invention;
[0016] FIG. 3 is a flow chart of sequentially representing the
image database sorting method in accordance with the present
invention; and
[0017] FIG. 4 provides a schematic diagram of describing a wavelet
transform process shown in FIG. 3.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0018] First of all, before mentioning an inventive image database
browsing and sorting method, there will be described the texture
extracted from a target static image in accordance with the present
invention.
[0019] The term "texture" represents a level of quality of an
image, i.e., coarseness of an object within the image. Therefore,
the image is expressed as a smooth image or a coarse image
according to a usual expression. However, the meaning of the level
of quality expressed as above is ambiguous and, thus, it can only
express a relative image quality for two different objects, i.e.,
relative coarseness or relative smoothness. For instance, the image
quality of two objects is expressed like an object A is coarser
than an object B. This relative notion can represent a relative
comparison but it cannot show a property for one object, i.e., an
absolute property.
[0020] However, the present invention provides the relative notion
of texture with an absolute level like cm that is a unit for
measuring height to thereby measure the texture of one object. As a
result, the texture introduced by the present invention becomes to
absolutely represent the coarseness or smoothness of each object in
an image without using a comparative object.
[0021] In accordance with the present invention, the texture is
described as a digit number. That is, the texture is expressed as a
number, e.g., 2.78 like the height is presented as 178 cm.
[0022] Hereinafter, with reference to the drawings, there will be
explained in detail a texture-based image database browsing and
sorting method in accordance with a preferred embodiment of the
present invention.
[0023] Referring to FIG. 1, there is shown a schematic diagram of
explaining an image database sorting method in accordance with the
present invention, which comprises a data image database (DB) 110,
a data image texture measuring unit 120 and a data image texture
database (DB) 130.
[0024] According to the image database sorting method, the data
image texture measuring unit 120 measures a texture of each image
stored in the data image DB 110, wherein a texture for each image
(static image, hereinafter referred to as a data image or an image)
is represented as one digit number obtained by computing the number
of edge pixel of an object within the static image. The measured
texture is stored in the data image texture DB 130 with the
original image (static image) Then, the measured texture values are
sorted according to a sorting order. That is, they are sorted in an
order their values are closer to a texture inputted in a next time.
For example, if the newly inputted texture is 10 and the previously
stored texture values are 15, 9, 7 and 20, as the sorting result,
the stored texture values are arranged in an order of 9, 7, 15 and
20.
[0025] Referring to FIG. 2, there is illustrated a schematic
diagram of showing an image database browsing method in accordance
with the present invention.
[0026] As shown in FIG. 2, the image DB browsing method further
includes a step of comparing and searching a data image and a query
image being a standard compared with the sorting method.
[0027] According to the sorting method described in FIG. 1, the
measured texture value of each data image (static image) is sorted
according to a sorting order and stored in the data image texture
DB 230.
[0028] Through the use of the stored texture values, the images
stored in the data image texture DB 230 can be browsed. As a
result, it is possible to more rapidly browse not apparent
properties but internal properties of images in a state of not
checking individual image data directly.
[0029] When browsing the properties of images, the texture of the
query image 240 is measured by computing the number of edge pixels
of an object within the query image 240 in the same manner as
described above. The measured texture of the query image 240 is
stored in a query image texture database (DB) 260.
[0030] Then, a texture comparison unit 270 performs the browsing or
sorting process by comparing the texture values of data images
close to each other by using the texture of the query image 240 as
a standard. Therefore, the internal properties of images can be
browsed without directly examining individual image data, wherein
the internal properties represent color, nature and objects in the
images.
[0031] Referring to FIG. 3, there is provided a flow chart of
sequentially showing an operation of a data image texture measuring
unit 220 in FIG. 2 in accordance with the present invention.
Through the operational sequence, the texture of the data image is
measured.
[0032] In step S310, an original image (static image) in the data
image DB is transformed through a Wavelet transform process to
thereby extract an image in a highest frequency region which
emphasizes a specific feature point included in the original
image.
[0033] At this time, the reason that the image in the highest
frequency region is extracted through the Wavelet transform process
is that a high frequency part represents edge information of
objects in the original image and a low frequency part shows
information for internal regions of the objects. Therefore, the low
frequency part can be used in an image compression process.
[0034] Then, since the extracted image in the highest frequency
region not only includes the edge information of the original image
but also noises therein, an image only including the edge
information is obtained by eliminating the noises through the use
of a Sobel operator in step S320.
[0035] In step S330, the number of edge pixels in the extracted
image is computed and the texture of the original image is obtained
by numerating the number of edge pixels determined through the
Sobel operator.
[0036] That is to say, if the number of edge pixels of an image A
is 10 and that of an image B 20, it means that the coarseness of
the image B is twice as much as that of the image A.
[0037] The Wavelet transform used in the present invention is a
technique which can resolve the data and functions in the original
image into frequency components different from each other and make
to examine results corresponding to each frequency and components
associated there to. Furthermore, this wavelet transform function
has various advantages required to the image compression, so that
it can be used as a standard compression coding technique in
MPEG-4.
[0038] The Sobel operator used in the present invention is a kind
of operator and is generally called a gradient operator. Through a
mask operation, it finds a vertical and horizontal edge of an
object as well as removing noises in images. It is defined as a
matrix consisting of:
[0039] Vertical components : [-1,0,1 : -2,0,2: -1,0,1]Horizontal
components : [-1,2,1: 0,0,0: 1,2,1]
[0040] Through the above matrix, the vertical and horizontal edge
information of the original image is obtained by performing a
product operation for the original image through the use of
3.times.3 mask.
[0041] The Wavelet transform technique including above features is
noticeable with reference to a drawing. Referring to FIG. 4, there
is provided a schematic diagram of describing a wavelet transform
process implemented in the present invention.
[0042] As illustrated in FIG. 4, the present invention executes an
n-dimensional wavelet transform using a high pass filter so as to
acquire edge information of objects in the static image.
[0043] In accordance with the present invention, the Wavelet
transform process is accomplished as dividing 4 frequency regions
411, 412, 413 and 414. Herein, a higher-lower filter region 413
represents a high frequency region to a low frequency region.
[0044] As before-mentioned, since the high frequency region
provides the edge information of the objects in the static image,
it is possible to perform the n-dimensional Wavelet transform
process by using the high frequency filter until developers obtain
desired images or edge information.
[0045] As described above, since the present invention measures
texture of static images by numerating the number of edge pixels of
objects in the static images, wherein the number of edge pixels is
computed by utilizing the Wavelet transform and the Sobel operator,
and sorts the measured textures in a corresponding image DB
according to a sorting order, the present invention can accomplish
fast and effective browsing based on contents of images and this
technique is applicable to the broadcasting data searching and
filtering which uses image data in a large amount.
[0046] While the present invention has been described with respect
to the particular embodiments, it will be apparent to those skilled
in the art that various changes and modifications may be made
without departing from the spirit and scope of the invention as
defined in the following claims.
* * * * *