U.S. patent application number 12/340522 was filed with the patent office on 2010-06-24 for system and method to selectively combine images.
This patent application is currently assigned to QUALCOMM Incorporated. Invention is credited to Kalin M. Atanassov, Hau Hwang, Hsiang-Tsun Li.
Application Number | 20100157079 12/340522 |
Document ID | / |
Family ID | 41718227 |
Filed Date | 2010-06-24 |
United States Patent
Application |
20100157079 |
Kind Code |
A1 |
Atanassov; Kalin M. ; et
al. |
June 24, 2010 |
SYSTEM AND METHOD TO SELECTIVELY COMBINE IMAGES
Abstract
Systems and methods to selectively combine images are disclosed.
In a particular embodiment, an apparatus includes a registration
circuit configured to generate a set of motion vector data based on
first image data corresponding to a first image and second image
data corresponding to a second image. The apparatus includes a
combination circuit to selectively combine the first image data and
adjusted second image data that corresponds to the second image
data adjusted according to the motion vector data. The apparatus
further includes a control circuit to control the combination
circuit to generate third image data.
Inventors: |
Atanassov; Kalin M.; (San
Diego, CA) ; Li; Hsiang-Tsun; (San Diego, CA)
; Hwang; Hau; (San Diego, CA) |
Correspondence
Address: |
QUALCOMM INCORPORATED
5775 MOREHOUSE DR.
SAN DIEGO
CA
92121
US
|
Assignee: |
QUALCOMM Incorporated
San Diego
CA
|
Family ID: |
41718227 |
Appl. No.: |
12/340522 |
Filed: |
December 19, 2008 |
Current U.S.
Class: |
348/222.1 ;
348/E5.022; 382/294; 455/556.2 |
Current CPC
Class: |
G06T 2207/10152
20130101; H04N 5/23232 20130101; G06T 3/0068 20130101; G06T
2207/10144 20130101; G06T 2207/10148 20130101; H04N 5/23267
20130101; G06T 5/50 20130101; H04N 5/23254 20130101; G06T
2207/20016 20130101 |
Class at
Publication: |
348/222.1 ;
382/294; 455/556.2; 348/E05.022 |
International
Class: |
H04N 5/228 20060101
H04N005/228; G06K 9/32 20060101 G06K009/32; H04M 1/00 20060101
H04M001/00 |
Claims
1. A method comprising: receiving first image data corresponding to
a first image and second image data corresponding to a second
image; adjusting the second image data by applying a set of motion
vector data corresponding to offsets between portions of the second
image data with respect to corresponding portions of the first
image data to produce adjusted second image data; and generating
third image data by selectively combining first values from the
first image data and second values from the adjusted second image
data at least partially based on comparing a first characteristic
of the first image data to a second characteristic of the second
image data.
2. The method of claim 1, wherein the selective combining includes
a coarse combining operation and a fine combining operation.
3. The method of claim 2, wherein the coarse combining operation is
performed on macroblocks of the adjusted second image data and
wherein the fine combining operation is performed on pixels of the
first image data and of the second adjusted image data.
4. The method of claim 2, wherein the coarse combining operation
includes selectively discarding one or more macroblocks of the
adjusted second image data when a difference between the one or
more macroblocks of the adjusted second image data and
corresponding macroblocks of the first image data exceeds a
selectable threshold value.
5. The method of claim 2, wherein the fine combining operation
includes filtering the first image data and the adjusted second
image data to compare filtered image characteristics, and wherein
the first values and the second values are selectively combined
based on the filtered image characteristics.
6. The method of claim 1, wherein the first characteristic and the
second characteristic are indicative of a focus condition, a
movement of an object within the first image and the second image,
or an exposure.
7. The method of claim 1, wherein the first values and the second
values are selectively combined at least partially based on
comparing a filtered portion of the first image data to a filtered
portion of the second image data.
8. The method of claim 1, wherein the first image data and the
second image data include successive frames captured at an image
sensor of a camera, wherein the first image data includes a first
set of macroblocks corresponding to the first image, wherein the
second image data includes a second set of macroblocks
corresponding to the second image, and wherein adjusting the second
image data includes applying the set of motion vector data to the
second set of macroblocks.
9. The method of claim 1, wherein the set of motion vector data is
determined via a hierarchical registration process.
10. The method of claim 9, wherein the hierarchical registration
process includes: determining a first alignment offset between a
first portion of the first image and a first portion of the second
image; and determining a second alignment offset between a second
portion of the first image and a second portion of the second image
based on the first alignment offset, wherein the second portion of
the first image is within the first portion of the first image.
11. The method of claim 1, wherein pixels within macroblocks of the
adjusted second image data are discarded when an image is adjusted
to compensate for hand jitter reduction.
12. The method of claim 1, wherein the third image data adjusts the
first image data and the second image data to compensate for
lateral chromatic aberrations.
13. A method comprising: determining a first set of motion vectors
corresponding to an offset between each block of a first set of
blocks of first image data and a corresponding block of a first set
of blocks of second image data; upsampling a motion vector of the
first set of motion vectors corresponding to a particular block of
the first set of blocks of the second image data to apply the
motion vector to a second set of blocks of the second image data,
the second set of blocks of the second image data included within
the particular block; determining a second set of motion vectors
corresponding to an offset between each of a second set of blocks
of the first image data and a corresponding block of the second set
of blocks of the second image data after applying the motion vector
to the second set of blocks; applying the second set of motion
vectors to the second image data to generate adjusted second image
data; and selectively combining the first image data and the
adjusted second image data to produce third image data.
14. The method of claim 13, wherein the third image data has at
least one of a greater depth of field than the first image data,
less noise than the first image data, or a greater dynamic range
than the first image data.
15. The method of claim 13, wherein the second set of motion
vectors is determined based on a selectable search range that is
indicated via a control input.
16. The method of claim 15, wherein the selectable search range
limits the offset between each of the second set of blocks of the
first image data and the corresponding block of the second set of
blocks of the second image data after applying the motion vector to
the second set of blocks.
17. An apparatus comprising: a registration circuit configured to
generate a set of motion vector data based on first image data
corresponding to a first image and second image data corresponding
to a second image; a combination circuit coupled to selectively
combine the first image data and adjusted second image data,
wherein the adjusted second image data corresponds to the second
image data after being adjusted according to the motion vector
data; and a control circuit to control the combination circuit to
generate third image data.
18. The apparatus of claim 17, wherein the registration circuit
includes a coarse registration circuit and a fine registration
circuit.
19. The apparatus of claim 17, wherein the combination circuit
includes a coarse combination circuit and a fine combination
circuit.
20. The apparatus of claim 17, wherein the registration circuit
comprises: a motion vector generation circuit having a first input
coupled to accept the first image data, a second input coupled to
accept the second image data, and a motion vector generation
output; a motion vector upsampling circuit having an input coupled
to accept the motion vector generation output and providing a
motion vector upsampling output; and a macro block motion vector
refining circuit having a first input coupled to accept the first
image data, a second input coupled to accept the second image data,
a third input coupled to accept the motion vector upsampling
output, and having a macro block motion vector output and a macro
block difference output.
21. The apparatus of claim 17, wherein the combination circuit
comprises: a macro block image alignment circuit having a first
input coupled to receive the first image data and a second input
coupled to receive the second image data, a third input coupled to
receive a macro block motion vector signal, an aligned first image
output and an aligned second image output; a block difference
discriminator circuit having a first input coupled to receive the
aligned first image output, a second input coupled to receive the
aligned second image output, a third input coupled to receive a
macro block difference signal, and having a first image difference
output and a second image difference output; and a mean pixel
difference circuit having a first input coupled to the block
difference discriminator circuit and configured to provide a pixel
difference output signal to indicate a mean pixel difference
result.
22. The apparatus of claim 21, further comprising: a first filter
coupled between the block difference discriminator circuit and the
mean pixel difference circuit having a first input coupled to
receive the first image difference output and an output providing a
filtered first image difference output; and a second filter coupled
between the block difference discriminator circuit and the mean
pixel difference circuit having a first input coupled to receive
the second image difference output and having an output to provide
a filtered second image difference output.
23. The apparatus of claim 17, further comprising: a camera lens;
and circuitry coupled to the camera lens to generate the first
image data and to generate the second image data.
24. The apparatus of claim 23, further comprising a processor
coupled to a wireless transceiver to send and receive data via an
antenna.
25. The apparatus of claim 23, wherein the apparatus is a personal
digital assistant ("PDA").
26. An apparatus comprising: a registration means for generating a
set of motion vector data based on first image data corresponding
to a first image and second image data corresponding to a second
image; a combination means for selectively combining the first
image data and adjusted second image data, wherein the adjusted
second image data corresponds to the second image data after being
adjusted according to the motion vector data; and a control means
for controlling the combination means to generate third image
data.
27. The apparatus of claim 26, further comprising: a camera lens;
and circuitry coupled to the camera lens to generate the first
image data and the second image data.
Description
I. FIELD
[0001] The present disclosure is generally related to a system and
method to selectively combine images.
II. DESCRIPTION OF RELATED ART
[0002] Advances in technology have resulted in smaller and more
powerful computing devices. For example, there currently exist a
variety of portable personal computing devices, including wireless
computing devices, such as portable wireless telephones, personal
digital assistants (PDAs), and paging devices that are small,
lightweight, and easily carried by users. More specifically,
portable wireless telephones, such as cellular telephones and
internet protocol (IP) telephones, can communicate voice and data
packets over wireless networks. Further, many such wireless
telephones include other types of devices that are incorporated
therein. For example, a wireless telephone can also include a
digital still camera, a digital video camera, a digital recorder,
and an audio file player. Such wireless telephones can process
executable instructions, including software applications, such as a
web browser application, that can be used to access the Internet.
As such, these wireless telephones can include significant
computing capabilities.
[0003] Digital signal processors (DSPs), image processors, and
other processing devices are frequently used in portable personal
computing devices that include digital cameras, or that display
image or video data captured by a digital camera. Such processing
devices can be utilized to provide video and audio functions, to
process received data such as captured image data, or to perform
other functions.
[0004] Captured image data may suffer from one or more issues such
as shifting errors due to hand jitter, movement of objects in the
image, overexposure, underexposure, poor focus in the near field or
far field, lateral chromatic aberrations, and geometric
distortions.
III. SUMMARY
[0005] Multiple images may be combined using a configurable image
processing architecture that performs registration and combination
of images to overcome issues that may occur in individual images. A
control unit may adjust operation of a hierarchical image
registration and a hierarchical image combination to enable various
effects of combining the input image data. For example, a first two
sets of image data may be combined for hand jitter reduction with
reduced ghosting, a next two sets of image data may be combined to
generate an enhanced depth of field image, and other sets of images
may be combined to generate a high dynamic range image.
[0006] In a particular embodiment, a method is disclosed that
includes receiving first image data corresponding to a first image
and second image data corresponding to a second image. The method
includes adjusting the second image data by applying a set of
motion vector data corresponding to offsets between portions of the
second image data with respect to corresponding portions of the
first image data to produce adjusted second image data. The method
further includes generating third image data by selectively
combining first values from the first image data and second values
from the adjusted second image data at least partially based on
comparing a first characteristic of the first image data to a
second characteristic of the second image data.
[0007] In another particular embodiment, a method is disclosed that
includes determining a first set of motion vectors corresponding to
an offset between each block of a first set of blocks of first
image data and a corresponding block of a first set of blocks of
second image data. The method also includes upsampling a motion
vector of the first set of motion vectors corresponding to a
particular block of the first set of blocks of the second image
data to apply the motion vector to a second set of blocks of the
second image data. The second set of blocks of the second image
data are included within the particular block. The method also
includes determining a second set of motion vectors corresponding
to an offset between each of a second set of blocks of the first
image data and a corresponding block of the second set of blocks of
the second image data after applying the motion vector to the
second set of blocks. The method also includes applying the second
set of motion vectors to the second image data to generate adjusted
second image data and selectively combining the first image data
and the adjusted second image data.
[0008] In another particular embodiment, an apparatus is disclosed.
The apparatus includes a registration circuit configured to
generate a set of motion vector data based on first image data
corresponding to a first image and second image data corresponding
to a second image. The apparatus also includes a combination
circuit to selectively combine the first image data and adjusted
second image data. The adjusted second image data corresponds to
the second image data that is adjusted according to the motion
vector data. The apparatus further includes a control circuit to
control the combination circuit to generate third image data.
[0009] In another particular embodiment, an apparatus is disclosed.
The apparatus includes a registration means for generating a set of
motion vector data based on first image data corresponding to a
first image and second image data corresponding to a second image.
The apparatus also includes a combination means for selectively
combining the first image data and adjusted second image data. The
adjusted second image data corresponds to the second image data
that is adjusted according to the motion vector data. The apparatus
further includes a control means for controlling the combination
means to generate third image data.
[0010] One particular advantage provided by embodiments of the
disclosed methods and apparatus is correction or improvement of
issues associated with images such as shifting due to hand jitter,
movement of objects in the image, overexposure, underexposure, poor
focus in the near field or far field, lateral chromatic
aberrations, and geometric distortions.
[0011] Other aspects, advantages, and features of the present
disclosure will become apparent after review of the entire
application, including the following sections: Brief Description of
the Drawings, Detailed Description, and the Claims.
IV. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] FIG. 1 is a block diagram of a particular illustrative
embodiment of a system including an image processing system having
an image combination module;
[0013] FIG. 2 is a block diagram illustrating a first embodiment of
a system including an image combination engine;
[0014] FIG. 3 is a block diagram illustrating a second embodiment
of a system including an image combination engine;
[0015] FIG. 4 is a block diagram illustrating a embodiment of a
system including an image combination circuit;
[0016] FIG. 5 is a diagram logically illustrating operation of an
embodiment of an image combination engine providing for correction
of hand jitter and for reducing object blur due to moving
objects;
[0017] FIG. 6 is a diagram logically illustrating operation of an
embodiment of an image combination engine to generate a high
dynamic range image;
[0018] FIG. 7 is a diagram logically illustrating operation of an
embodiment of an image combination engine providing for depth of
field enhancement;
[0019] FIG. 8 illustrates lateral chromatic aberrations;
[0020] FIG. 9 illustrates a first embodiment of geometric
distortions of an image;
[0021] FIG. 10 illustrates a second embodiment of geometric
distortions of an image;
[0022] FIG. 11 is a diagram logically illustrating operation of an
embodiment of an image combination engine providing for correction
of geometric distortions of an image;
[0023] FIG. 12 is a flow diagram of a method of selectively
combining images;
[0024] FIG. 13 is a flow diagram of a method of performing a
hierarchical combination process;
[0025] FIG. 14 is a flow diagram of a method of performing an image
improvement process; and
[0026] FIG. 15 is a block diagram of a portable electronic device
including an image combination module.
V. DETAILED DESCRIPTION
[0027] FIG. 1 is a block diagram of a particular illustrative
embodiment of a system including an image processing system having
an image combination module. The system 100 includes an image
capture device 101 coupled to an image processing system 130. The
image processing system 130 is coupled to an image storage device
140. The image processing system 130 is configured to receive image
data 109 from the image capture device 101 and to combine multiple
images providing correction or improvement of issues associated
with the multiple images, such as shifting due to hand jitter,
movement of objects in the image, overexposure, underexposure, poor
focus in the near field or far field, lateral chromatic
aberrations, and geometric distortions. Generally, the system 100
may be implemented in an electronic device that is configured to
perform real-time image processing using relatively limited
processing resources.
[0028] In a particular embodiment, the image capture device 101 is
a camera, such as a video camera or a still camera. In other
embodiments, the image capture device 101 may be a camera embodied
in a cellular telephone, personal digital assistant (PDA) or the
like. The image capture device 101 includes a lens 102 that is
responsive to a focusing module 104 and to an exposure module 106.
A sensor 108 is coupled to receive light via the lens 102 and to
generate the image data 109 in response to an image received via
the lens 102. The focusing module 104 may be responsive to the
sensor 108 and may be adapted to automatically control focusing of
the lens 102. The exposure module 106 may also be responsive to the
sensor 108 and may be adapted to control an exposure of the image.
In a particular embodiment, the sensor 108 includes multiple
detectors, or pixel wells, that are arranged so that adjacent
detectors detect different colors of light. For example, received
light may be filtered so that each detector receives red, green, or
blue incoming light.
[0029] The image capture device 101 is coupled to provide the image
data 109 to the image processing system 130. The image processing
system 130 includes a demosaic module 110 to perform a demosaic
operation on image data 109 received from the image capture device
101. A color correction module 112 is configured to perform color
correction on demosaiced image data. A gamma module 114 is
configured to generate gamma corrected image data from data
received from the color correction module 112. A color conversion
module 116 is coupled to perform a color space conversion to the
gamma corrected image data. The image processing system 130 also
includes an image combination module 118 that is configured to
combine multiple images, as is discussed with respect to FIGS.
2-14. A compress and store module 120 is coupled to receive an
output of the image combination module 118 and to store compressed
output data at the image storage device 140. The image storage
device 140 may include any type of storage medium, such as one or
more display buffers, registers, caches, flash memory elements,
hard disks, any other storage device, or any combination
thereof.
[0030] As was discussed, in a particular embodiment, the sensor 108
includes multiple detectors that detect different colors of light,
such as red, green and blue (RGB). Thus, images may be received in
the RGB color space. In a particular embodiment, the images may be
converted to other color spaces such as the "YCbCr" color space by
the color conversion module 116. The YCbCr color space is an
example of a color space where images are represented by a luma (or
brightness) component (the Y component in the YCbCr color space)
and chroma components (the Cb and Cr components in the YCbCr color
space.) In the YCbCr color space, Cb is blue minus luma (B-Y) and
Cr is red minus luma (R-Y).
[0031] During operation, the image combination module 118 may
correct or improve images corresponding to the input image data
109. For example, corrections or improvements may be made to images
to compensate for issues associated with the captured images, such
as shifting due to hand jitter, movement of objects in the image,
overexposure, underexposure, poor focus in the near field or far
field, lateral chromatic aberrations, and geometric
distortions.
[0032] Although in the particular embodiment illustrated in FIG. 1
the image combination nodule 118 follows the color conversion
module 116 in the image processing pipeline, in other embodiments
the image combination module 118 may be implemented at other
locations within the image processing pipeline. In addition,
although the image capture device 102 is illustrated as having a
single sensor 108, in other embodiments the image capture device
102 may have multiple sensors. For example, the image capture
device 102 may have two or more sensors configured to perform
multiple concurrent image captures of a particular scene under
various conditions, such as using different exposure settings or
different focus settings.
[0033] FIG. 2 illustrates a system 200 having an image combination
engine 205 with a first image input 201 to receive first image data
corresponding to a first image. In a particular embodiment, the
system 200 may be included in the image combination module 118 of
FIG. 1. For example, a first image may be captured by the camera
lens 102 and provided at the first image input 201 to the image
combination engine 205 through the sensor 108, the demosaic module
110, the color correction module 112, the gamma module 114 and the
color conversion module 116.
[0034] The image combination engine 205 further includes a second
image input 202 to receive second image data corresponding to a
second image. In a particular embodiment, the second image may be
captured by the camera lens 102 and provided at the second image
input 202 to the image combination engine 205 through the sensor
108, the demosaic module 110, the color correction module 112, the
gamma module 114 and the color conversion module 116.
[0035] The first and second images are combined by the image
combination engine 205. For example, the image combination engine
205 may operate according to embodiments described with respect to
FIGS. 3-14.
[0036] The image combination engine 205 is configured to generate a
combined image output 206. In a particular embodiment, the image
combination engine 205 generates the combined image output 206 by
selectively combining first values from the first image input 201
and adjusted second values from the second image input 202 at least
partially based on comparing a first characteristic of a first
image to a second characteristic of a second image. For example,
when a region of the first image is out of focus but the
corresponding region of the second image is in focus, the image
combination engine 205 may select pixel values for the combined
image output 206 corresponding to the region from the second image.
Other examples of characteristics that may be compared include
contrast of the images, deviations between the luminance components
of the images, and filtered characteristics of the images (e.g.
low-pass filtered data, high-pass filtered data). In a particular
embodiment, the combined image output 206 may be coupled to an
image storage device such as image storage device 140.
[0037] In other embodiments, the image combination engine 205 may
include additional image inputs. In addition or alternatively, the
combined image output 206 may be coupled to the first image input
201 providing for the ability to combine the output of the image
combination engine 205 with additional image inputs received on the
second image input 202 to iteratively combine the output image with
additional input images. For example, three sets of image data may
be combined to form a single image by applying a first set of image
data to the first image input 201 and a second set of image data to
the second image input 202, and by applying the resulting combined
image to the first image input 201 and the third set of image data
to the third image input 202, resulting in a combined image of all
three sets of image data.
[0038] FIG. 3 illustrates a system 300 having an image combination
engine 305. In the system 300, the image combination engine 305
includes a registration circuit 321 which registers separate image
inputs, a combination circuit 323 which combines separate image
inputs, and a control circuit 325 that controls the registration
circuit 321 and the combination circuit 323. In an illustrative
embodiment, the system 300 may be included in the image combination
module 118 of FIG. 1 or the image combination engine 205 of FIG.
2.
[0039] In a particular embodiment, the registration circuit 321
includes a first luma input 311, a second luma input 313, and an
output 322 coupled to the combination circuit 323. The registration
circuit 321 is configured to determine differences between data
from the first luma input 311 and the second luma input 313 and to
provide offset data to the combination circuit 323 at the output
322. In a particular embodiment, the first luma input 311 is the Y
component data for a first image coded in the YCbCr color space and
the second luma input 313 is the Y component data for a second
image coded in the YCbCr color space. As illustrated in FIG. 3, the
registration circuit 321 performs image registration using only the
luma components from the images to be registered. In other
embodiments, the registration circuit 321 may use other components
of the image data in addition to, or in place of, the luma
component to perform registration.
[0040] The combination circuit 323 includes a first image input
315, a second image input 317, an input to accept registration data
from the registration circuit 321 and a combined image output 306.
In a particular embodiment, the first image input 315 receives data
for a first image coded in the YCbCr color space and the second
image input 317 receives data for a second image coded in the YCbCr
color space. The combination circuit 323 is configured to
selectively combine a first image and a second image that has been
adjusted based on differences detected by the registration circuit
321 (i.e., adjusted to align with the first image data). The
registration circuit 321 is configured to be responsive to the
input 326 to operate under control of the control circuit 325 and
the combination circuit 323 is configured to be responsive to the
input 324 to operate under control of the control circuit 325.
[0041] In a particular embodiment, the image combination circuit
323 generates the combined image output 306 by selectively
combining first values from the first image input 315 and second
values from the second image input 317 at least partially based on
comparing a first characteristic of a first image to a second
characteristic of a second image. The characteristics may include
focus, contrast, variance, or frequency spectrum, as illustrative
examples. For example, the image combination circuit 323 may
combine regions of the first image input 315 and the second image
input 317 based on which image input has better focus in the
region, which image input has better contrast in the region, how
much variance is detected between the two image inputs in the
region after the registration, or other characteristics to improve
a quality of the combined image output 306. The image combination
circuit 323 may receive input 324 from the control circuit 325
indicating selected characteristics, evaluate the respective images
on a region-by-region or even pixel-by-pixel basis, and generate
the combined image based on a selective combination of regions or
pixels of the input images based on the evaluated
characteristics.
[0042] Referring to FIG. 4, a system to selectively combine
multiple images is depicted and generally designated 400. In a
particular embodiment, the system 400 may be included in the image
combination module 118 of FIG. 1, the image combination engine 205
of FIG. 2, the image combination engine 305 of FIG. 3, or any
combination thereof.
[0043] The system 400 includes a hierarchical registration circuit
420 that is coupled to a hierarchical combination circuit 460. The
hierarchical registration circuit 420 and the hierarchical
combination circuit 460 are coupled to an application specific
control circuit 432. The application specific control circuit 432
and the hierarchical combination circuit 460 are also coupled to a
weighting table 434.
[0044] The hierarchical registration circuit 420 is configured to
receive first image luma data 402 corresponding to a first image
and second image luma data 404 corresponding to a second image and
to perform a registration process on the first image luma data 402
and the second image luma data 404 using a coarse registration
circuit 422 and a fine registration circuit 424. The hierarchical
registration circuit 420 is configured to generate a fine set of
motion vectors 429 that indicate detected offsets between
corresponding portions of the first image luma data 402 and the
second image luma data 404. In a particular embodiment, the fine
set of motion vectors 429 include magnitude and direction data to
align images that may be misaligned due to camera movement, image
movement, or both. As illustrated, the hierarchical registration
circuit 420 operates on image luma data for computational
efficiency. However, in other embodiments, the hierarchical
registration circuit 420 may operate using other types of image
data, such as chroma component data, red data, blue data, or green
data, or any combination thereof, in addition to or in place of
luma data.
[0045] In a particular embodiment, the coarse registration circuit
422 includes a motion vector generation circuit 426. The motion
vector generation circuit 426 may be configured to partition each
of the first image luma data 402 and the second image luma data 404
into blocks to perform a coarse registration process between the
blocks. For example, the motion vector generation circuit 426 may
logically divide each of the first image luma data 402 and the
second image luma data 404 into a 3.times.3 set of overlapping
blocks and may use a projection of the overlapping blocks to
generate a coarse set of motion vectors 427 that can be applied to
align each of the blocks of the second image luma data 404 to a
corresponding block of the first image luma data 402. In other
embodiments, any number of blocks may be used, and some or all of
the blocks may be non-overlapping blocks.
[0046] The fine registration circuit 424 is configured to receive
the coarse set of motion vectors 427 and to generate a fine set of
motion vectors 429. In a particular embodiment, the fine
registration circuit 424 includes a motion vector upsampling
circuit 428 coupled to a macro block refining circuit 430. The
motion vector upsampling circuit 428 may receive and upsample the
coarse set of motion vectors 427 to generate motion vectors having
a finer granularity than the coarse set of motion vectors 427. To
illustrate, the image luma data 402 and 404 may be configured as
M.times.N arrays of macro blocks, where each macro block
corresponds to a sixteen-pixel-by sixteen-pixel region of an image.
The motion vector upsampling circuit 428 may generate a M.times.N
set of motion vectors that applies the corresponding motion vector
of the coarse set of motion vectors 427 to each macro block.
[0047] In a particular embodiment, the macro block motion vector
refining circuit 430 is coupled to receive the upsampled set of
motion vectors 427 and the image luma data 402 and 404 and to
generate a refined set of motion vectors 429. For example, the
macro block motion vector refining circuit 430 may be configured to
apply each motion vector of the upsampled set of motion vectors 427
to its corresponding macro block of the second image data to
coarsely align the macro block of the second image data with a
corresponding macro block of the first image data. The macro block
motion vector refining circuit 430 may search a region of the first
image data 402 around the coarsely aligned macro block to determine
a more accurate alignment of the coarsely aligned macro block to
the first image data 402. The search region may be selected based
on a search range control signal 435 received from the application
specific control circuit 432. The refined set of motion vectors 429
may indicate vector data corresponding to the more accurate
alignment of each macro block to enable a macro block-by-macro
block registration of the first image luma data 402 and the second
image luma data 404.
[0048] The macro block motion vector refining circuit 430 may
determine the refined set of motion vectors 429 by performing an
algorithm that selects a lowest calculated mean square error (MSE)
or other norm among multiple possible MSEs for each motion vector.
For example, for a particular macro block of the second image luma
data 404, multiple possible alignments of the macro block with the
first image luma data 402 may be considered, and the alignment that
results in a lowest computed MSE is selected for the refined set of
motion vectors 429. The mean square error determined for each
macroblock may be provided to the hierarchical combination circuit
460 as motion vector (MV) means square difference data 431.
[0049] In a particular embodiment, the hierarchical combination
circuit 460 is configured to combine first image data 406 and
second image data 408 using a coarse combination circuit 462 and a
fine combination circuit 464. The first image data 406 may include
the first luma data 402 for the first image and also includes
chroma data for the first image as YCbCr image data. The second
image data 408 may include the second luma data 404 for the second
image and chroma data for the second image as YCbCr data.
[0050] In a particular embodiment, the coarse combination circuit
462 includes a macro block image alignment circuit 466 and a block
MSE difference discriminator circuit 468. The macro block image
alignment circuit 466 may be configured to apply the refined set of
motion vectors 429 to the second image data 408 to generate image
data for the second image that is aligned to the first image data.
For example, the macro block image alignment circuit 466 may be
configured to combine pixel values in the second image when macro
blocks are determined to overlap, or to interpolate pixel values
where macro blocks are realigned to result in a region of the
second image data that is not within any macro blocks. The macro
block image alignment circuit 466 may provide the first image data
406 and the aligned image data for the second image to the block
MSE difference discriminator circuit 468.
[0051] In a particular embodiment, the block MSE difference
discriminator circuit 468 is configured to perform a coarse
combination process on the data received from the macro block image
alignment circuit 466. In particular, the block MSE difference
discriminator circuit 468 may eliminate macro blocks of the aligned
image data for the second image that do not sufficiently match the
first image data 406. For example, the MV MS difference data 431
for each macro block may be compared against a threshold value.
When the MS difference exceeds the threshold value for a particular
macro block, the particular macro block is determined to be too
different between the first image data 406 and the aligned image
data for the second image, and thus the image data should not be
combined for the particular macro block.
[0052] For example, where a moving object appears in a first macro
block in the first image data 406 (but not in the first macro block
in the aligned image data for the second image) and the moving
object appears in a second macro block in the aligned image data
for the second image (but not in the second macro block of the
first image data 406), the first macro block may be determined to
be non-combinable between the first and second images, and the
second macro block may determined to be non-combinable between the
first and second images, due to the corresponding mean square error
differences. The block MSE difference discriminator circuit 468 may
be configured to remove each non-combinable macro block from the
aligned second image data so that only the pixel values for the
macro block from the first image data 406 are used. For example,
the pixel values for the macro block may be copied from the first
image data 406 to replace the pixel values in the corresponding
macroblock of the aligned image data for the second image.
[0053] As illustrated, the block MSE difference discriminator
circuit 468 is responsive to the application specific control
circuit 432. For example, the application specific control circuit
432 may provide a threshold control signal 437 that indicates a
threshold difference to be used to compare MSE differences between
macroblocks of the first image data and the aligned image data for
the second image. The block MSE difference discriminator circuit
468 may output two sets of image data to the fine combination
circuit 464, including image data corresponding to the first image
and image data corresponding to the second image following the
coarse combination process.
[0054] The fine combination circuit 464 is configured to receive
first and second image data that has been registered and coarsely
aligned, and to perform a fine combination process to generate
output image data 480. In a particular embodiment, the fine
combination circuit 464 includes a first filter 470 and a second
filter 472 coupled to a mean pixel MS difference circuit 474. The
fine combination circuit 464 also includes an image combining
circuit 476 coupled to the mean pixel MS difference circuit 474 and
to the weighting table 434.
[0055] The received data for the first image may be processed by
the first filter 470, and the filtered data for the first image is
provided to the mean pixel MS difference circuit 474. The received
data for the second image may be processed by the second filter
472, and the filtered data for the second image is provided to the
mean pixel MS difference circuit 474. The filters 470 and 472 may
be responsive to the application specific control circuit 432. For
example, the filters 470 and 472 may receive a response control
signal 439 from the application specific control circuit 432 that
indicates a filter response characteristic, such as a low-pass
response, a high-pass response, a bandpass response, any other
filter response, or any combination thereof. The filters 470 and
472 may include a 3.times.3 kernel, or any other size kernel. In a
particular embodiment, the filters 470 and 472 have a kernel size
responsive to the application specific control circuit 432.
[0056] The mean pixel MS difference circuit 474 may be configured
to receive the filtered data corresponding to each image and to
perform a pixel-by-pixel signed mean square difference operation.
The difference operation may generate a signed value indicating a
difference between the filtered data of the first image and the
filtered data of the second image, for each particular pixel, using
each of the luma and chroma values for the particular pixel. The
mean pixel MS difference circuit 474 may be configured to provide
the difference data to the image combining circuit 476 as a mean
pixel difference result.
[0057] The image combining circuit 476 may be configured to
receive, for each pixel, a difference value from the mean pixel MS
difference circuit 474 and to determine a pixel value of each pixel
in the output image data 480. For example, the received difference
value for a particular pixel may be provided as a lookup operation
at the weighting table 434. A result of the lookup operation may
determine whether the pixel value in the output image data 480 has
a value from the first image data received from the coarse
combination circuit 462, a value from the second image data
received from the coarse combination circuit 462, or a combination
thereof.
[0058] The weighting table 434 may include data indicating a first
weight to be applied to a pixel value of the first image data and a
second weight to be applied to a pixel value of the second image
data. The weighting table 434 may provide an output value "W"
having a range of approximately 0 to 1 that corresponds to a weight
to be applied to the first image data, and a value 1-W that
corresponds to a weight to be applied to the second image data. The
weighting table may be responsive to a table control signal 433
from the application specific control circuit 432.
[0059] During operation, the application specific control circuit
432 may determine one or more control parameters to control an
image registration and combination process at the system 400. For
example, the application specific control circuit 432 may select a
value of the search range control signal 435 to indicate an
aggressiveness of the macro block registration, the threshold
control signal 433 to indicate an amount of acceptable difference
for macroblock combination, the response control signal 439 to
indicate a type of filtering to be performed, and the table control
signal 433 to indicate a weighting of how the images are to be
combined based on a filtered pixel difference between the
images.
[0060] Although the system 400 is illustrated as including hardware
circuits configured to perform specific processes, in other
embodiments one or more components of the system 400 may be
performed by a processor executing processor instructions. For
example, one or more of the functions performed by the circuits
420, 422, 424, 426, 428, 430, 432, 434, 460, 462, 464, 466, 468,
470, 474, or 476 may be performed by an image processor, digital
signal processor (DSP), or general purpose processor that has been
programmed to perform one or more of the functions or general
algorithms described above. In other embodiments, one or more of
the circuits 420, 422, 424, 426, 428, 430, 432, 434, 460, 462, 464,
466, 468, 470, 474, or 476 may be replaced by components included
in hardware, firmware, a processor executing computer readable
instructions, or any combination thereof.
[0061] Particular embodiments illustrating image combining are
discussed in connection with FIGS. 5-11 and are useful for
understanding operation of the image combination circuit 400 of
FIG. 4 as may be implemented in these particular embodiments.
[0062] Referring to FIG. 5, a diagram is provided logically
illustrating operation of an embodiment of an image combination
engine providing for correction of hand jitter and reducing object
blur due to moving objects. For purposes of illustration in FIG. 5,
a moving object 502 is represented by a circle and a portion of an
image shifted due to jitter 504 is illustrated by a triangle. A
first column 521 generally represents a processing path of a first
image and a second column 523 generally represents a processing
path of a second image.
[0063] In an illustrative embodiment, the data flow 501 of the
first and second images to remove hand jitter may be performed in
the image combination module 118 of FIG. 1, the image combination
engine 205 of FIG. 2, the image combination engine 305 of FIG. 3,
the image combination circuit 400 of FIG. 4, or any combination
thereof. In a second illustrative embodiment, the data flow 501 of
the first and second images to remove hand jitter may be performed
in accordance with the method of selective combination of images
1200 of FIG. 12, the hierarchical registration process 1300 of FIG.
13, the image improvement process 1400 of FIG. 14, or any
combination thereof.
[0064] Initially, the first image and the second image are provided
as inputs to a coarse registration process resulting in a coarse
registration 512 and a fine registration process resulting in a
fine registration 514. The coarse registration process and the fine
registration process are configured to determine differences
between the first image and the second image. As illustrated, the
coarse registration 512 may subdivide each set of image data into
portions such as a first portion 530 and may determine an offset
between the first portion of the first image and the first portion
of the second image. The fine registration 514 may further
subdivide each portion, such as into macroblocks that correspond to
sixteen-pixel-by-sixteen-pixel areas of the images, illustrated as
a second portion 532 that is within the first portion 530. The fine
registration 516 may determine an offset between the second portion
of the first image and the second portion of the second image, and
may represent the offset via a motion vector, such as a motion
vector of the fine set of motion vectors 429 of FIG. 4.
[0065] Blocks of the second image are aligned with blocks of the
first image based on the registration of the images to produce a
coarse combination block alignment 516.
[0066] A coarse combination block difference process detects the
moving object 502 (represented by the circle) based on the number
of pixels the object 502 shifted between the first image and the
second image to produce a coarse combination block difference 518.
The number of pixels an object must shift before it is considered
to be a moving object may vary from application to application and,
in certain embodiments, is controlled by an application-specific
registration control module, such as the application specific
registration control module 431 of FIG. 4 or the control circuit
325 of FIG. 3. After the moving object is detected, the moving
object is removed from the second image by replacing the block with
the moving object in the second image with the corresponding block
of the first image, which may substantially reduce "ghosting" due
to moving objects when combining images. The shift due to hand
jitter 504, illustrated as the triangle, is adjusted by the coarse
registration process and the fine registration process so that the
images can be combined on a pixel-by-pixel basis. For example, the
fine combination process may average values of the first and second
registered images for reduced image error and reduced noise. A
resulting combined image provides correction or improvement of
issues associated with the images such as shifting due to hand
jitter and ghosting due to movement of objects in the image, with
less noise than either of the first or second image data.
[0067] As an illustrative, non-limiting example the data flow 501
may be performed at the system 400 of FIG. 4 in a hand jitter
reduction with object blur reduction mode controlled by the
application specific control circuit (ASCC) 432. The ASCC 432 may
instruct the hierarchical registration circuit 420 to use a large
search range to detect object motion via the search range control
signal 435. The ASCC 432 may instruct the coarse combination
circuit 462 via the threshold signal 437 to not combine blocks that
are far off from each other, such as by setting the threshold
according to a largest expected or measured noise value plus a
predetermined margin. The ASCC 432 may configure the filters 470
and 472 of the fine combination circuit 464 via the response signal
439 to operate in an all-pass mode for each plane (e.g., Y, Cb, and
Cr) to enable comparison based on the filtered image
characteristics. The ASCC 432 may configure the weighting table 434
via the table signal 433 to provide a weighting function that
causes the fine combination circuit 464 to generate a pixel value
of the output image data 480 using an average of the pixel's value
in the first image data and the second image data received from the
coarse combination circuit 462 when the magnitude of the mean pixel
MS difference of the filtered image data is less than a selected
amount, and to use only the pixel's value in the first image data
otherwise.
[0068] Referring to FIG. 6, a diagram is shown logically
illustrating operation of an embodiment of an image combination to
generate a high dynamic range image. A first column 621 represents
a processing path of an image taken with a relatively short
exposure time resulting in part of the image being properly exposed
and represented as a proper exposure portion 602 and part of the
image being under-exposed and represented as an under-exposed
portion 604. A second column 623 represents a processing path of an
image taken with a relatively long exposure time resulting in part
of the image being over-exposed represented as an over-exposed
portion 606 and part of the image being properly exposed and
represented as a properly exposed portion 608.
[0069] In an illustrative embodiment, the data flow 601 of the
first and second images to provide a high dynamic range (HDR) image
may be performed in the image combination module 118 of FIG. 1, the
image combination engine 205 of FIG. 2, the image combination
engine 305 of FIG. 3, the image combination circuit 400 of FIG. 4,
or any combination thereof. In a second illustrative embodiment,
the data flow 601 of the first and second images to provide a HDR
image may be performed in accordance with the method of selective
combination of images 1200 of FIG. 12, the hierarchical
registration process 1300 of FIG. 13, the image improvement process
1400 of FIG. 14, or any combination thereof.
[0070] Initially, the first image in the first processing path 621
and the second image in the second processing path 623 are provided
as inputs to a coarse registration process that provides a coarse
registration 612 and a fine registration process that provides a
fine registration 614. The coarse registration process and the fine
registration process determine differences between the first image
and the second image.
[0071] Blocks of the second image are aligned with blocks of the
first image based on the registration of the images to produce a
coarse combination 616. The coarse combination 616 further has
blocks of the registered images removed that do not adequately
match, such as due to object motion in the image, as described in
FIG. 5.
[0072] A fine combination process combines the proper exposure
portion 602 of the first image with the proper exposure portion 608
of second image, on a pixel-by-pixel basis, resulting in a fine
combination having a properly exposed HDR image. In certain
embodiments, other image enhancement functions may be carried out
using the fine combination process.
[0073] As an illustrative, non-limiting example the data flow 601
may be performed at the system 400 of FIG. 4 in a high dynamic
range mode controlled by the application specific control circuit
(ASCC) 432. The ASCC 432 may instruct the hierarchical registration
circuit 420 via the search range control signal 435 to not rely on
fine motion vector estimation, such as to use a very small or zero
search range. The ASCC 432 may instruct the coarse combination
circuit 462 via the threshold signal 437 to use a very high
threshold or to disable discarding blocks. The ASCC 432 may
configure the filters 470 and 472 of the fine combination circuit
464 via the response signal 439 to set a luma filter for the
reference image to average and everything else to zero to enable
comparison based on the filtered image characteristics. The ASCC
432 may configure the weighting table 434 via the table signal 433
to provide a weighting function that causes the fine combination
circuit 464 to generate a pixel value of the output image data 480
using the pixel's value in the first image data when the mean pixel
MS difference of the filtered image data is less than a first
amount, using the pixel's value in the second image data when the
mean pixel MS difference of the filtered image data is greater than
a second amount, and a smooth transition to an average of the
pixel's value in the first image data and the second image data
received from the coarse combination circuit 462 when the magnitude
of the mean pixel MS difference of the filtered image data is
between the first amount and the second amount.
[0074] FIG. 7 is a diagram logically illustrating operation of an
embodiment of an image combination engine providing for depth of
field enhancement. As illustrated by FIG. 7, a first image in a
first processing path 721 includes a near field portion 702 which
is focused and a far field portion 704 which is blurry. A second
image in a second processing path 723 includes a near field portion
706 which is blurry and a far field portion 708 which is
focused.
[0075] In an illustrative embodiment, the data flow 701 of the
first and second images to provide depth of field enhancement may
be performed in the image combination module 118 of FIG. 1, the
image combination engine 205 of FIG. 2, the image combination
engine 305 of FIG. 3, the image combination circuit 400 of FIG. 4,
or any combination thereof. In a second illustrative embodiment,
the data flow 701 of the first and second images to provide depth
of field enhancement may be performed in accordance with the method
of selective combination of images 1200 of FIG. 12, the
hierarchical registration process 1300 of FIG. 13, the image
improvement process 1400 of FIG. 14, or any combination
thereof.
[0076] Initially, the first image and the second image are provided
as inputs to a coarse registration process producing a coarse
registration image 712 and a fine registration process producing a
fine registration image 714. The coarse registration image 712 and
the fine registration image 714 are configured to determine
differences between the first image and the second image. After
registration, blocks of the second image are aligned with blocks of
the first image based on the registration of the images by the
coarse combination process to produce a coarse combination image
716. The coarse combination process further removes non-matching
blocks due to moving objects in one of the registered images.
[0077] A fine combination process combines the focused near field
portion 702 of the first image with the focused far field portion
708 of the second image on a pixel-by-pixel basis resulting in a
focused combined image to produce a fine combination image 718. In
certain embodiments, other image enhancement functions may be
carried out using the fine combination process. An image with an
enhanced depth of field is provided.
[0078] As an illustrative, non-limiting example the data flow 701
may be performed at the system 400 of FIG. 4 in an enhanced depth
of field mode controlled by the application specific control
circuit (ASCC) 432. The ASCC 432 may instruct the hierarchical
registration circuit 420 via the search range control signal 435 to
use a large search range to enable object move detection. The ASCC
432 may instruct the coarse combination circuit 462 via the
threshold signal 437 to not combine blocks that are far off from
each other, such as by setting the threshold according to a largest
expected or measured noise value plus a predetermined margin. The
ASCC 432 may configure the filters 470 and 472 of the fine
combination circuit 464 via the response signal 439 to have a high
pass filter response for luma data and zero for chroma data to
enable comparison based on the filtered image characteristics. The
ASCC 432 may configure the weighting table 434 via the table signal
433 to provide a weighting function that causes the fine
combination circuit 464 to generate a pixel value of the output
image data 480 using the pixel's value in the first image data when
the mean pixel MS difference of the filtered image data is less
than zero, and transitioning to use the pixel's value in the second
image data when the mean pixel MS difference of the filtered image
data is greater than zero.
[0079] Turning to FIG. 8, lateral chromatic aberrations are
illustrated. As illustrated by FIG. 8, incoming light is depicted
as planes 803, 805 and 807 incident on a lens. Due to refractive
properties of the lens, each detected color of the incident light
may have a slightly different field of view, illustrated as fields
of view 813, 815 and 817, respectively, as received at a sensor
array. In a particular embodiment, registration of the three color
planes generated using a sensor output, such as red, green, and
blue color planes, can compensate for the different fields of
view.
[0080] FIG. 9 and FIG. 10 illustrate images 902 and 1002
respectively having geometric distortions. Geometric distortions
may be caused by a number of factors including lens distortions and
other distortions in the mechanical, optical and electrical
components of an imaging system. In a particular embodiment,
resampling of the geometrically distorted images can correct for
geometric distortions.
[0081] FIG. 11 is a block diagram logically illustrating operation
of an embodiment of an image combination engine providing for
correction of geometric distortions of an image. In FIG. 11, a
single image with geometric distortions 1104 is provided as an
input to a coarse registration process to produce a coarse
registration image 1112. The coarse registration process may use a
set of coarse registration vectors that may be predefined and based
on known geometric distortions caused by mechanical, optical, and
electrical components of the imaging system to make a coarse
resampling of the image 1104.
[0082] A fine registration process may then use a set of fine
registration vectors that may be predefined and based on known
geometric distortions to make a fine resampling of the image 1104
to produce a fine registration image 1114. The resampled image 1115
is then provided to a course combination process producing a coarse
combination image 1116 and a fine combination module producing a
fine combination image 1118. In a particular embodiment, the image
1104 may also be combined with other images for other corrections
providing a corrected image 1120.
[0083] In an illustrative embodiment, the data flow 1101 of an
image to provide correction of geometric distortions of the image
may be performed in the image combination module 118 of FIG. 1, the
image combination engine 205 of FIG. 2, the image combination
engine 305 of FIG. 3, the system 400 of FIG. 4, or any combination
thereof. In a second illustrative embodiment, the data flow 1101 of
an image to provide correction of geometric distortions of the
image may be performed in accordance with the method of selective
combination of images 1200 of FIG. 12, the hierarchical
registration process 1300 of FIG. 13, the image improvement process
1400 of FIG. 14, or any combination thereof.
[0084] Turning to FIG. 12, a method of selective combination of
images is generally shown at 1200. In an illustrative embodiment,
the method 1200 may be performed by the image combining module 118
of FIG. 1, the image combination engine 205 of FIG. 2, the image
combination engine 305 of FIG. 3, the system 400 of FIG. 4, or any
combination thereof. First image data corresponding to a first
image and second image data corresponding to a second image is
received, at 1202. In a particular embodiment, the first image data
and the second image data may be received in a color space having a
luma component and a chroma component. In another particular
embodiment, the first image data and second image data may be
received in the YCbCr color space. A set of motion vector data may
be calculated corresponding to offsets between portions of the
second image data with respect to corresponding portions of the
first image data, at 1204. For example, the set of motion vector
data may be calculated using a coarse registration circuit to
generate a first set of motion vectors to roughly align the image
data and a fine registration circuit to generate a finer set of
motion vectors based on the first set of motion vectors.
[0085] The second image data is adjusted by applying the set of
motion vector data which, as described, corresponds to offsets
between portions of the second image data with respect to
corresponding portions of the first image data, at 1206. As a
result, the second image data is adjusted to more closely align to
the first image data, to compensate for movement of the camera or
for movement of objects in the image between capturing the first
image data and the second image data. Third image data is generated
by selectively combining first values from the first image data and
second values from the adjusted second image data at least
partially based on comparing a first characteristic of the first
image data to a second characteristic of the second image data, at
1208. Examples of characteristics that may be compared include
focus of the images, contrast of the images, deviations between the
luminance components of the images, and filtered characteristics of
the images (e.g. low-pass filtered data, high-pass filtered
data).
[0086] In a particular embodiment, the selective combining includes
a coarse combining operation and a fine combining operation. The
course combining operation may be performed on macroblocks of the
adjusted second image data and the fine combining operation may be
performed on pixels of the first image data and the second adjusted
image data. The coarse combining operation may include selectively
discarding one or more macroblocks of the adjusted second image
data when a difference between the one or more macroblocks of the
adjusted second image data and corresponding macroblocks of the
first image data exceeds a selectable threshold value. For example
the selectable threshold may be indicated by the threshold control
signal 437 provided by the application specific control circuit 432
of FIG. 4. In another particular embodiment, the fine combining
operation includes filtering the first image data and the adjusted
second image data to compare filtered image characteristics, and
the first values and second values are selectively combined based
on the filtered image characteristics. The filtered image
characteristics may be compared at the mean pixel difference
circuit 474 of FIG. 4 and used to query the weighting table 434 to
generate weights for a weighted combination of the first image data
and the adjusted second image data.
[0087] In a particular embodiment, the first characteristics of the
first image data and the second characteristic of the second image
data are indicative of the focus condition of the images (e.g., in
focus or out of focus), a movement of an object within the first
image and the second image, or exposure of the images. In addition,
in a particular embodiment, the values of the first and second
image data may be combined based on comparing a filtered portion of
the first image data with a filtered portion of the second image
data. For example, the filtered portion of the first image data may
be compared with the filtered portion of the second image data at
the mean pixel difference circuit 474 of FIG. 4, and a result of
the comparison may be used to determine a relative weight to be
applied to the values of the first image data and the values of the
second image data. The first image data and the second image data
may be, for example, successive frames captured at an image sensor
of a camera.
[0088] In certain embodiments, pixels or blocks of the first or
second image data may be discarded and replaced with pixels or
blocks from the other of the first or second image data when
certain conditions are met. For example, pixels within macroblocks
of the adjusted second image may be discarded when an image is
adjusted for hand jitter or detection of motion. Pixels in the
first image or in the adjusted second image may be discarded for
depth of field enhancement. The resultant third image data may be
enhanced over the first image or the second image for hand jitter,
movement, depth of field enhancement, lateral chromatic aberrations
and geometric distortions.
[0089] In a particular embodiment, as illustrated by FIG. 13, the
set of motion vector data is determined based on a hierarchical
registration process 1300. The hierarchical registration process
1300 may be performed by the image combining module 118 of
[0090] FIG. 1, the image combination engine 205 of FIG. 2, the
image combination engine 305 of FIG. 3, the system 400 of FIG. 4,
or any combination thereof. The hierarchical structure process
provides the ability to adjust resolution versus robustness
operating points and reduces computational demand. The hierarchical
registration process 1300 includes determining a first alignment
offset between a first portion of the first image and a first
portion of the second image, at 1302, and determining a second
alignment offset between a second portion of the first image and a
second portion of the second image based on the first alignment
offset, where the second portion of the first image is within the
first portion of the first image, at 1304. In one particular
embodiment, the first portion of the first image is one of nine
divisions in a 3.times.3 matrix of the first image and the second
portion of the first image corresponds to a
sixteen-pixel-by-sixteen-pixel area of the first image.
[0091] FIG. 14 depicts a flowchart that illustrates an embodiment
of an image improvement method 1400. In an illustrative embodiment,
the image improvement method 1400 can be performed by the image
combining module 118 of FIG. 1, the image combination engine 205 of
FIG. 2, the image combination engine 305 of FIG. 3, the system 400
of FIG. 4, or any combination thereof. A first set of motion
vectors is determined corresponding to an offset between each block
of a first set of blocks of first image data and a corresponding
block of a first set of blocks of second image data, at 1402. For
example, the first set of motion vectors may be the coarse set of
motion vectors 427 of FIG. 4.
[0092] A motion vector of the first set of motion vectors
corresponding to a particular block of the first set of blocks of
the second image data is upsampled to apply the motion vector to a
second set of blocks of the second image data, the second set of
blocks of the second image data included within the particular
block, at 1404. For example, the motion vector upsampling circuit
428 of FIG. 4 may upsample the motion vector of a particular block
of a 3.times.3 set of overlapping blocks to a set of macro blocks
within the particular block.
[0093] A second set of motion vectors corresponding to an offset
between each of a second set of blocks of the first image data and
a corresponding block of the second set of blocks of the second
image data after applying the motion vector of the first set of
motion vectors are determined, at 1406. For example, the second set
of motion vectors may be the fine set of motion vectors 429
generated by the macro block refining circuit 430 of FIG. 4. The
second set of motion vectors are applied to the second image data
to generate adjusted second image data, at 1408.
[0094] In a particular embodiment, the second set of motion vectors
are determined based on a selectable search range that is indicated
via a control input, such as the input 326 of FIG. 3 or search
range control signal 435 that is received at the macro block
refining circuit 430 of FIG. 4. The selectable search range may
limit the offset between each of the second set of blocks of the
first image data and the corresponding block of the second set of
blocks of the second image data after applying the motion vector of
the first set of motion vectors.
[0095] Portions of the first image data and the adjusted second
image data are selectively combined, at 1410. For example, the
first image data and the second image data may be combined on a
region-by-region or a pixel-by-pixel basis, or both. To illustrate,
the first image data and the adjusted second image data may be
combined by the hierarchical combination circuit 460 of FIG. 4.
[0096] In a particular embodiment, the selective combining of the
first image data and the adjusted second image data produces third
image data. The third image data may have a greater depth of field
than the first image data, less noise than the first image data, or
a greater dynamic resolution than the first image data.
[0097] FIG. 15 is a block diagram of particular embodiment of a
system including a image combination module. The system 1500 may be
implemented in a portable electronic device and includes a signal
processor 1510, such as a digital signal processor (DSP), coupled
to a memory 1532. The system 1500 includes an image combination
module 1564. In an illustrative example, the image combination
module 1564 includes any of the systems of FIGS. 1-5, operates in
accordance with any of the methods of FIGS. 12-14, or any
combination thereof. The image combination module 1564 may be
incorporated into the signal processor 1510 or may be a separate
device.
[0098] A camera interface 1568 is coupled to the signal processor
1510 and is also coupled to a camera, such as a camera 1570. The
camera 1570 may be a video camera or a still image camera or may
implement both functionalities. A display controller 1526 is
coupled to the signal processor 1510 and to a display device 1528.
A coder/decoder (CODEC) 1534 can also be coupled to the signal
processor 1510. A speaker 1536 and a microphone 1538 can be coupled
to the CODEC 1534. A wireless interface 1540 can be coupled to the
signal processor 1510 and to a wireless antenna 1542.
[0099] In a particular embodiment, the signal processor 1510
includes the image combination module 1564 that is adapted to
determine a first set of motion vectors corresponding to an offset
between each block of a first set of blocks of first image data and
a corresponding block of a first set of blocks of second image
data. The image combination module 1564 may be adapted to upsample
a motion vector of the first set of motion vectors corresponding to
a particular block of the first set of blocks of the second image
data to apply the motion vector to a second set of blocks of the
second image data. The second set of blocks of the second image
data are included within the particular block. The image
combination module 1564 may be adapted to determine a second set of
motion vectors corresponding to an offset between each of a second
set of blocks of the first image data and a corresponding block of
the second set of blocks of the second image data after applying
the motion vector of the first set of motion vectors. The image
combination module 1564 may be adapted to apply the second set of
motion vectors to the second image data to generate adjusted second
image data. The image combination module 1564 may be adapted to
selectively combine portions of the first image data and the
adjusted second image data to produce third image data.
[0100] For example, the image combination module 1564 may include
the hierarchal registration circuit 420 of FIG. 4 including the
motion vector upsampling circuit 428 to upsample a motion vector of
the coarse set of motion vectors 427. The upsampled motion vector
may correspond to a particular block of a 3.times.3 grid of blocks,
and may be upsampled to each macro block within the particular
block. The image combination module 1564 may also include the
hierarchical combination circuit 460 of FIG. 4 to selectively
combine the image data, such as at the coarse combination circuit
462 and at the fine combination circuit 464.
[0101] The signal processor 1510 may also be adapted to generate
image data that has been processed by the image combination module
1564. The processed image data may include video data from the
video camera 1570, image data from a wireless transmission via the
wireless interface 1540, or from other sources such as an external
device coupled via a universal serial bus (USB) interface (not
shown), as illustrative, non-limiting examples.
[0102] The display controller 1526 is configured to receive the
processed image data and to provide the processed image data to the
display device 1528. In addition, the memory 1532 may be configured
to receive and to store the processed image data, and the wireless
interface 1540 may be configured to receive the processed image
data for transmission via the antenna 1542.
[0103] In a particular embodiment, the signal processor 1510, the
display controller 1526, the memory 1532, the CODEC 1534, the
wireless interface 1540, and the camera interface 1568 are included
in a system-in-package or system-on-chip device 1522. In a
particular embodiment, an input device 1530 and a power supply 1544
are coupled to the system-on-chip device 1522. Moreover, in a
particular embodiment, as illustrated in FIG. 15, the display
device 1528, the input device 1530, the speaker 1536, the
microphone 1538, the wireless antenna 1542, the video camera 1570,
and the power supply 1544 are external to the system-on-chip device
1522. However, each of the display device 1528, the input device
1530, the speaker 1536, the microphone 15315, the wireless antenna
1542, the video camera 1570, and the power supply 1544 can be
coupled to a component of the system-on-chip device 1522, such as
an interface or a controller.
[0104] In a particular embodiment, the system 1500 may function as
a personal digital assistant ("PDA"), a cellular telephone or
similar device. The system 1500 may be adapted to provide for user
controllable input, such as through input device 1530, and may
include a control circuit to control the control image combination
module 1564 and to receive the user controllable input.
[0105] Those of skill would further appreciate that the various
illustrative logical blocks, configurations, modules, circuits, and
algorithm steps described in connection with the embodiments
disclosed herein may be implemented as electronic hardware,
computer software, or combinations of both. To clearly illustrate
this interchangeability of hardware and software, various
illustrative components, blocks, configurations, modules, circuits,
and steps have been described above generally in terms of their
functionality. Whether such functionality is implemented as
hardware or software depends upon the particular application and
design constraints imposed on the overall system. Skilled artisans
may implement the described functionality in varying ways for each
particular application, but such implementation decisions should
not be interpreted as causing a departure from the scope of the
present disclosure.
[0106] The steps of a method or algorithm described in connection
with the embodiments disclosed herein may be embodied directly in
hardware, in a software module executed by a processor, or in a
combination of the two. A software module may reside in random
access memory (RAM), flash memory, read-only memory (ROM),
programmable read-only memory (PROM), erasable programmable
read-only memory (EPROM), electrically erasable programmable
read-only memory (EEPROM), registers, hard disk, a removable disk,
a compact disc read-only memory (CD-ROM), or any other form of
storage medium known in the art. An exemplary storage medium is
coupled to the processor such that the processor can read
information from, and write information to, the storage medium. In
the alternative, the storage medium may be integral to the
processor. The processor and the storage medium may reside in an
application-specific integrated circuit (ASIC). The ASIC may reside
in a computing device or a user terminal. In the alternative, the
processor and the storage medium may reside as discrete components
in a computing device or user terminal.
[0107] The previous description of the disclosed embodiments is
provided to enable any person skilled in the art to make or use the
disclosed embodiments. Various modifications to these embodiments
will be readily apparent to those skilled in the art, and the
principles defined herein may be applied to other embodiments
without departing from the scope of the disclosure. Thus, the
present disclosure is not intended to be limited to the embodiments
shown herein but is to be accorded the widest scope possible
consistent with the principles and novel features as defined by the
following claims.
* * * * *