U.S. patent application number 14/202798 was filed with the patent office on 2014-07-10 for method and apparatus for blending images.
This patent application is currently assigned to SAMSUNG ELECTRONICS CO., LTD.. The applicant listed for this patent is Samsung Electronics Co., Ltd.. Invention is credited to Won-Hee CHOE, Jae-Hyun KWON, Seong-Deok LEE, Hyun-Hwa OH, Sung-Chan PARK.
Application Number | 20140193097 14/202798 |
Document ID | / |
Family ID | 43756669 |
Filed Date | 2014-07-10 |
United States Patent
Application |
20140193097 |
Kind Code |
A1 |
PARK; Sung-Chan ; et
al. |
July 10, 2014 |
METHOD AND APPARATUS FOR BLENDING IMAGES
Abstract
A method and apparatus may include aligning a reference image
and an input image through ROI-based image alignment; detecting an
ROI associated with an incompletely aligned region in the aligned
image; and blending the aligned reference image and input image by
using a plurality of weights which are attenuated according to a
spatial distance from a boundary of the detected ROI. In addition,
the method and apparatus may include aligning a reference image and
an input image through image alignment based on an n.sup.th region
of interest (ROI) which is detected at a previous stage; detecting
another ROI associated with an incompletely aligned region in the
aligned image; and blending the aligned image in the other ROI and
an (n+1).sup.th blended image, which is input from a next stage, in
a region outside the other ROI, and outputting the last blended
image to the previous stage.
Inventors: |
PARK; Sung-Chan; (Suwon-si,
KR) ; OH; Hyun-Hwa; (Hwaseong-si, KR) ; KWON;
Jae-Hyun; (Hwaseong-si, KR) ; LEE; Seong-Deok;
(Seongnam-si, KR) ; CHOE; Won-Hee; (Gyeongju-si,
KR) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Samsung Electronics Co., Ltd. |
Suwon -si |
|
KR |
|
|
Assignee: |
SAMSUNG ELECTRONICS CO.,
LTD.
Suwon-si
KR
|
Family ID: |
43756669 |
Appl. No.: |
14/202798 |
Filed: |
March 10, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
12886114 |
Sep 20, 2010 |
8699822 |
|
|
14202798 |
|
|
|
|
Current U.S.
Class: |
382/284 |
Current CPC
Class: |
G06T 2207/20208
20130101; G06T 2207/20221 20130101; G06T 5/50 20130101 |
Class at
Publication: |
382/284 |
International
Class: |
G06T 5/50 20060101
G06T005/50 |
Foreign Application Data
Date |
Code |
Application Number |
Sep 23, 2009 |
KR |
10-2009-0089875 |
Claims
1. A method for blending images, comprising: aligning a reference
image and an input image through region of interest (ROI)-based
image alignment; detecting an ROI associated with an incompletely
aligned region in the aligned image; and blending the aligned input
image and reference image by using a plurality of weights which are
attenuated according to a spatial distance from a boundary of the
detected ROI.
2. The method of claim 1, wherein blending the aligned input image
and reference image comprises blending two images by taking the
reference image in the ROI and the input image in a region outside
the ROI.
3. The method of claim 1, wherein detecting the ROI comprises
detecting a ghost.
4. The method of claim 1, wherein the weights are more greatly
attenuated at an edge of an image than at other portions of the
image.
5. The method of claim 1, wherein the weights are attenuated
depending on a difference in brightness level which occurs in the
boundary of the ROI.
6. A method for blending images, comprising: aligning a reference
image and an input image through image alignment based on an
n.sup.th region of interest (ROI) which is detected at a previous
stage; detecting another ROI associated with an incompletely
aligned region in the aligned image; blending the aligned image in
the other ROI and an (n+1).sup.th blended image, which is input
from a next stage, in a region outside the other ROI; and
outputting the last blended image to the previous stage.
7. The method of claim 6, wherein detecting the other ROI comprises
detecting a ghost.
8. The method of claim 6, further comprising outputting an
(n+1).sup.th ROI, which is an overlapped portion between the
n.sup.th ROI and the other ROI, to a next stage.
9. The method of claim 6, wherein blending the aligned image and
the (n+1).sup.th blended image comprises blending the two images by
using a plurality of weights which are attenuated according to a
spatial distance from a boundary of the other ROI.
10. The method of claim 9, wherein the weights are more greatly
attenuated at an edge of an image than at the other portions of the
image.
11. The method of claim 9, wherein the weights are attenuated
depending on a difference in brightness level which occurs in the
boundary of the other ROI.
12. An image blender, comprising: an image aligning unit configured
to align a reference image and an input image through ROI-based
image alignment; an ROI detecting unit configured to detect an ROI
associated with an incompletely aligned region by detecting motion
in an aligned image output from the image alignment unit; a weight
calculating unit configured to calculate a plurality of weights
which are attenuated depending on a spatial distance from a
boundary of the ROI detected by the ROI detecting unit; and an
image blending unit comprising a processor, the image blending unit
configured to blend an aligned image output from the image aligning
unit and a reference image according to the weights which are
calculated by the weight calculating unit.
13. The image blender of claim 12, wherein the ROI detecting unit
comprises a ghost detecting unit configured to detect a ghost.
14. An image blender, comprising: an image aligning unit configured
to align a reference image and an input image through image
alignment, based on an n.sup.th ROI which is detected at a previous
stage; an ROI detecting unit configured to detect another ROI
associated with an incompletely aligned region, based on motion
detection in an aligned image output from the image aligning unit;
and an image blending unit comprising a processor, the image
blending unit configured to: blend the aligned input image in the
other ROI and an (n+1).sup.th blended image, which is input from a
next stage, in a region outside the other ROI; and output the last
blended image to the previous stage.
15. The image blender of claim 14, wherein the ROI detecting unit
comprises a ghost detecting unit configured to detect ghosts, based
on motion detection in the aligned image output from the image
aligning unit.
16. The image blender of claim 14, wherein the ROI detecting unit
is further configured to output an (n+1).sup.th ROI, which is an
overlapped portion between the n.sup.th ROI and the other ROI, to a
next stage.
17. The image blender of claim 14, further comprising a weight
calculating unit configured to calculate a plurality of weights,
the image blending unit being further configured to blend the two
images using the plurality of weights, the plurality of weights
being attenuated depending on a spatial distance from a boundary of
the other ROI.
18. An image blending apparatus, comprising: is a plurality of
image blenders, at least two of the plurality of image blenders
comprising: an image aligning unit configured to align a reference
image and an input image through image alignment, based on an
n.sup.th ROI which is detected at a previous stage; an ROI
detecting unit configured to detect another ROI associated with an
incompletely aligned region, based on motion detection in an
aligned image output from the image aligning unit; and an image
blending unit comprising a processor, the image blending unit
configured to: blend the aligned image in the other ROI and an
(n+1).sup.th blended image, which is input from a next stage, in a
region outside the other ROI; and output the last blended image to
the previous stage.
19. The image blending apparatus of claim 18, wherein the plurality
of image blenders are cascaded.
20. The image blending apparatus of claim 19, wherein an initial
ROI of the first stage image blender is set to a low-brightness
region, or a high-brightness region.
21. A non-transitory computer-readable information storage medium
storing a program for causing a computer to implement the method of
claim 1.
22. A non-transitory computer-readable information storage medium
storing a program for causing a computer to implement the method of
claim 6.
Description
CROSS REFERENCE TO RELATED APPLICATION(S)
[0001] This application is a Continuation Application of U.S.
patent application Ser. No. 12/886,114, filed Sep. 20, 2010, which
claims the benefit under 35 U.S.C. .sctn.119(a) of Korean Patent
Application No. 10-2009-0089875, filed on Sep. 23, 2009, the entire
disclosures of which is incorporated herein by reference for all
purposes.
BACKGROUND
[0002] 1. Field
[0003] The following description relates to a method and apparatus
for blending a plurality of images.
[0004] 2. Description of the Related Art
[0005] Image blending involves compiling multiple images to achieve
a desired effect. Image blending may be applied to high dynamic
range imaging (HDRI) for merging a number of standard photographs
captured using exposure bracketing into an HDR image. Further, the
image blending may be applied to panoramic imaging for merging a
number of partly overlapped photographs into a wide-angle view or
representation of a physical space. In addition, the image blending
may be applied to photomontages for making a composite photograph
by cutting and joining a number of other photographs. For example,
the photomontage may be used in making a composite photograph by
cutting and joining desired parts of other photographs. Such
techniques may be applied to various products, such as analog
cameras, digital cameras, and camcorders, i.e., in the image
capture device. The techniques may be further used in image
processing using computers, i.e., outside of the image capture
device.
[0006] For such image blending, multiple images need to be aligned
and a difference in brightness level at a stitched boundary of two
different regions needs to fall within a predetermined limit.
Further, there should be no artifacts in the boundary between the
two different regions so that the blended image appears naturally,
i.e., so that the output appears that the images were originally
combined with no visible seams or inconsistencies.
[0007] However, it is very difficult for images with noise or with
a portion in active motion to be blended without decreasing dynamic
range thereof or causing motion artifacts. Using an optical flow
technique to solve the motion artifacts may result in a decreased
matching level and an increased complexity of calculation for
images with considerable motion.
SUMMARY
[0008] In one general aspect, there is provided a method for
blending images, including: aligning a reference image and an input
image through region of interest (ROI)-based image alignment,
detecting an ROI associated with an incompletely aligned region in
the aligned image, and blending the aligned input image and
reference image by using a plurality of weights which are
attenuated according to a spatial distance from a boundary of the
detected ROI.
[0009] The method may further include that blending the aligned
input image and reference image includes blending two images by
taking the reference image in the ROI and the input image in a
region outside the ROI.
[0010] The method may further include that detecting the ROI
includes detecting a ghost.
[0011] The method may further include that the weights are more
greatly attenuated at an edge of an image than at other portions of
the image.
[0012] The method may further include that the weights are
attenuated depending on a difference in brightness level which
occurs in the boundary of the ROI.
[0013] In another general aspect, there is provided a method for
blending images, including: aligning a reference image and an input
image through image alignment based on an n.sup.th region of
interest (ROI) which is detected at a previous stage, detecting
another ROI associated with an incompletely aligned region in the
aligned image, blending the aligned input image in the other ROI
and an (n+1).sup.th blended image, which is input from a next
stage, in a region outside the other ROI, and outputting the last
blended image to the previous stage.
[0014] The method may further include that detecting the other ROI
includes detecting a ghost.
[0015] The method may further include outputting an (n+1).sup.th
ROI, which is an overlapped portion between the n.sup.th ROI and
the other ROI, to a next stage.
[0016] The method may further include that blending the aligned
input image and the (n+1).sup.th blended image includes blending
the two images by using a plurality of weights which are attenuated
according to a spatial distance from a boundary of the other
ROI.
[0017] The method may further include that the weights are more
greatly attenuated at an edge of an image than at the other
portions of the image.
[0018] The method may further include that the weights are
attenuated depending on a difference in brightness level which
occurs in the boundary of the other ROI.
[0019] In another general aspect, there is provided an image
blender, including: an image aligning unit configured to align a
reference image and an input image through ROI-based image
alignment, an ROI detecting unit configured to detect an ROI
associated with an incompletely aligned region by detecting motion
in an aligned image output from the image alignment unit, a weight
calculating unit configured to calculate a plurality of weights
which are attenuated depending on a spatial distance from a
boundary of the ROI detected by the ROI detecting unit, and an
image blending unit configured to blend an aligned image output
from the image aligning unit and a reference image according to the
weights which are calculated by the weight calculating unit.
[0020] The image blender may further include that the ROI detecting
unit includes a ghost detecting unit configured to detect a
ghost.
[0021] In another general aspect, there is provided an image
blender, including: an image aligning unit configured to align a
reference image and an input image through image alignment, based
on an n.sup.th ROI which is detected at a previous stage, an ROI
detecting unit configured to detect another ROI associated with an
incompletely aligned region, based on motion detection in an
aligned image output from the image aligning unit, and an image
blending unit configured to: blend the aligned input image in the
other ROI and an (n+1).sup.th blended image, which is input from a
next stage, in a region outside the other ROI, and output the last
blended image to the previous stage.
[0022] The image blender may further include that the ROI detecting
unit includes a ghost detecting unit configured to detect ghosts,
based on motion detection in the aligned image output from the
image aligning unit.
[0023] The image blender may further include that the ROI detecting
unit is further configured to output an (n+1).sup.th ROI, which is
an overlapped portion between the n.sup.th ROI and the other ROI,
to a next stage.
[0024] The image blender may further include a weight calculating
unit configured to calculate a plurality of weights, the image
blending unit being further configured to blend the two images
using the plurality of weights, the plurality of weights being
attenuated depending on a spatial distance from a boundary of the
other ROI.
[0025] In another general aspect, there is provided an image
blending apparatus, including: a plurality of image blenders, at
least two of the plurality of image blenders including: an image
aligning unit configured to align a reference image and an input
image through image alignment, based on an n.sup.th ROI which is
detected at a previous stage, an ROI detecting unit configured to
detect another ROI associated with an incompletely aligned region,
based on motion detection in an aligned image output from the image
aligning unit, and an image blending unit configured to: blend the
aligned reference image in the other ROI and an (n+1).sup.th
blended image, which is input from a previous stage, in a region
outside the other ROI, and output the last blended image to the
previous stage.
[0026] The image blending apparatus may further include that the
plurality of image blenders are cascaded.
[0027] The image blending apparatus may further include that an
initial ROI of the first offstage image blenders is set to a
low-brightness region, a high-brightness region.
[0028] A computer-readable information storage medium may store a
program for causing a computer to implement the above methods.
[0029] Other features and aspects may be apparent from the
following detailed description, the drawings, and the claims.
BRIEF DESCRIPTION OF THE DRAWINGS
[0030] FIG. 1 is a block diagram of an image blender according to
an example embodiment.
[0031] FIG. 2 is a block diagram of a cascaded structure of two
image blenders according to an example embodiment.
[0032] FIG. 3 is a block diagram of an image blender according to
another example embodiment.
[0033] FIG. 4 is a block diagram of a cascaded structure of a
plurality of image blenders according to an example embodiment.
[0034] FIG. 5 is a block diagram of an image blender according to
still another example embodiment.
[0035] FIG. 6 is an example method for blending images.
[0036] FIG. 7 is another example method for blending images.
[0037] Throughout the drawings and the detailed description, unless
otherwise described, the same drawing reference numerals will be
understood to refer to the same elements, features, and structures.
The relative size and depiction of these elements may be
exaggerated for clarity, illustration, and convenience.
DETAILED DESCRIPTION
[0038] The following detailed description is provided to assist the
reader in gaining a comprehensive understanding of the methods,
apparatuses, and/or systems described herein. Accordingly, various
changes, modifications, and equivalents of the systems, apparatuses
and/or methods described herein will be suggested to those of
ordinary skill in the art. The progression of processing steps
and/or operations described is an example; however, the sequence of
steps and/or operations is not limited to that set forth herein and
may be changed as is known in the art, with the exception of steps
and/or operations necessarily occurring in a certain order. Also,
descriptions of well-known functions and constructions may be
omitted for increased clarity and conciseness.
[0039] FIG. 1 is a block diagram of an image blender according to
an example embodiment.
[0040] The image blender may include an image aligning unit 110, a
region of interest (ROI) detecting unit 130, a weight calculating
unit 150, and an image blending unit 170. The image aligning unit
110 may align a reference image and an input image according to
ROI-based image alignment. The ROI detecting unit 130 may detect an
ROI associated with an incompletely aligned region by detecting
motion in an aligned image output from the image aligning unit 110.
The weight calculating unit 150 may calculate a plurality of
weights which are attenuated depending on a spatial distance from a
boundary of the ROI detected by the ROI detecting unit 130. The
image blending unit 170 may blend the reference image and the
aligned image output from the image aligning unit 110 according to
the weights calculated by the weight calculating unit 150.
[0041] The units shown in FIG. 1 may be individually separated or
incorporated into a single entity. For example, signal processors
or the image blender may be implemented as a field programmable
gate array (FPGA) or gate array (GA). In another embodiment, the
units may be implemented as a computer and program code. In still
another embodiment, the units may be implemented as a digital
signal processor and a set of commands executed by the digital
signal processor in an analog digital camera, a digital camera or a
camcorder.
[0042] An initial ROI may be a low-brightness region with a
brightness lower than a predetermined value in an example
embodiment. However, the initial ROI may be a high-brightness
region.
[0043] The image aligning unit 110 may align a reference image
I.sup.REF and an input image I.sup.IN through ROI (R.sub.0)-based
image alignment. Examples of the ROI-based image alignment may
include a hierarchical pyramid structure of alignment, which has
been proposed by the present inventor, and other known alignment
schemes. The output image may be well aligned in the initial ROI,
which may be a low brightness region, but may be incompletely
aligned in the other region.
[0044] The ROI detecting unit 130 may detect another ROI associated
with the incompletely aligned region in the aligned image. For
example, the ROI may be a ghost region where motion is detected. In
the current example, the ROI detecting unit 130 may be implemented
as a ghost detecting unit. The term "ghost" refers to a phenomenon
in which images overlap due to motion of a subject in the images
when the images are merged or stitched.
[0045] The weight calculating unit 150 may calculate a plurality of
weights which are attenuated according to a spatial distance from a
boundary of the ROI detected by the ROI detecting unit 130. A ROI
may be a space between two squares. By applying different weights
depending on a spatial distance from the boundary of the ROI, a
difference in brightness level may be improved and motion artifacts
may be reduced in the boundary of the ROI.
[0046] In an example, a detected ROI may be set to have a weight of
"1", and a region farther from the boundary of the ROI may be set
to have a lower weight. In other words, the weights may be
attenuated according to a spatial distance from the boundary of the
ROI.
[0047] In another example, where an ROI is denoted by "R" and a
distance range parameter is denoted by "a", a kernel "K" is defined
as follows:
K i = [ 1 - ax i , ax i < 1 0 , otherwise ] ( 1 ) K N = exp ( -
1 2 ax 2 ) ( 2 ) ##EQU00001##
[0048] In one example, each pixel may have a different region by
adjusting the distance range parameter.
[0049] Using the kernel K, the weights may be attenuated according
to a spatial distance from the center as follows:
W ( x ) = max k .di-elect cons. I K ( k - x ) .delta. R ( k ) ,
.delta. R ( k ) = { 1 , k .di-elect cons. R 0 , k R ( 3 )
##EQU00002##
[0050] The image blending unit 170 may blend two images according
to the weights calculated by the weight calculating unit 150. That
is, the image blending unit 170 may use the above-mentioned weight
function W(x) to blend a reference image I.sup.REF(x) in an ROI and
an input image I.sub.k(x) in another region as follows:
I out ( x ) = W ( x ) I REF ( x ) + k ( 1 - W ( x ) ) I k ( x ) ( 4
) ##EQU00003##
[0051] In one example, the reference image may be taken in the ROI
and the input image may be taken in the region other than the ROI
when the two images are blended. In the embodiment shown, as a well
aligned image may be taken in the ROI showing no ghost, and a
reference image as the input image may be taken in the other
region, motion artifacts may be reduced.
[0052] However, the weights may be attenuated in a gradual curve.
Further, the weight values may not need to be symmetrically
attenuated with respect to the boundary of the ROI. In one
embodiment, the weights may be set to be more significantly
attenuated at an edge than at other portions. This may prevent the
corresponding region of the image from deteriorating in sharpness
due to feathering in the weights
[0053] FIG. 2 is a block diagram of an image blender according to
another example embodiment. FIG. 2 is a block diagram of a cascaded
structure of two image blenders according to an example
embodiment.
[0054] As shown in FIG. 2, the cascaded image blender may include a
first image blender 100 and a second image blender 300 which may be
cascaded. An overlapped portion between an ROI detected by an ROI
detecting unit 130 of the first image blender 100 and an ROI
detected by an image aligning unit 130 of the first image blender
100 may be set to an ROI of an image aligning unit 310 of the
second image blender 300. Further, an output image from an image
blending unit 370 of the second image blender 300 may be input to
an image blending unit 170 of the first image blender 100.
[0055] That is, in the first image blender 100, the image aligning
unit 110 may align an initial ROI, and the ROI detecting unit 130
may detect a misaligned region. In the second image blender 300,
the image aligning unit 310 may align images in the narrowed ROI,
and the ROI detecting unit 330 may detect a misaligned ghost
region. The image blending unit 370 may blend the initial ROI in an
ROI G.sub.2 and a well-aligned I.sub.2 image in the other region,
and may output the blended image. The image blending unit 170 of
the first image blender 100 may blend a well-aligned I.sub.1 image
in an ROI G.sub.1 detected by the ROI detecting unit 130 and the
output image of the second image blender 300 in the other region,
and may output the blended image.
[0056] The image aligning unit 110 of the first image blender 100
may process the ROI into a well-aligned low-brightness region using
a hierarchical pyramid structure of alignment. The ROI G.sub.1 may
then be calculated by detecting a ghost between the aligned image
I.sub.1 and the reference image.
[0057] The image I.sub.2 may then be calculated by performing
realignment upon a region R1, which may be an overlap region
between the ghost-detected image G.sub.1 and the low-brightness ROI
of the first image blender. The ROI G.sub.2 may then be calculated
by detecting a ghost. Then, a final image may be output by
performing ROI-based blending of the I.sub.2 image and the
reference image I.sup.REF in the ROI G.sub.2, and performing
ROI-based blending of the I.sub.1 image and the previously blended
image in the ROI G.sub.1.
[0058] This process may be represented in two stages when ROI-based
deghosting and ROI-based image alignment are considered as being a
single module. For example, an image I.sub.k may be calculated by
receiving an ROI R.sub.k-1 from the previous stage and performing
image alignment for the ROI R.sub.k-1, and an image G.sub.k may be
calculated by detecting a ghost using the images I.sub.k and
I.sub.REF.
[0059] In this manner, I.sub.k and G.sub.k may be calculated for
each level k from low to high levels. The following operation may
then be performed level by level from high to low levels.
[0060] That is, a blending weight W.sub.k may be calculated using
G.sub.k, the image I.sub.k and the image I.sub.k+1.sup.F may be
input and blended, and the resultant image I.sub.k.sup.F may be
transferred to a next lower level. Finally, the image I.sub.1.sup.F
may be output as a last image at level 1. In this case, the input
R.sub.0 at level 1 may become an initial ROI, and the output may
become the last blended image. If the highest level is K, the input
image I.sub.k+1.sup.F may become I.sub.REF at the level K.
[0061] This may be represented in three stages as shown in the
example of FIG. 4. The alignment may be performed once, as shown in
the example of FIG. 1. By iterating the same process in this
manner, it is possible to increasingly extend the alignment
region.
[0062] FIG. 3 is a block diagram of an image blender according to
another example embodiment. The image blender may be implemented in
a program with a single sub-routine. Alternatively, the image
blender may be implemented in a single chip or semiconductor IP
(Intellectual Property) core. Such modules may be cascaded to
provide a better image blender.
[0063] An ROI detection or ghost detection process may be performed
by setting an initial ROI, performing image alignment for the
initial ROI, detecting a ghost in the ROI, and resetting another
ROI. Accordingly, by iterating the image alignment and ghost
detection process for each ROI, the image I.sub.k and the ghost
region G.sub.k, which may be aligned based on each ROI whenever the
process is iterated, may be acquired. This is expressed as
follows:
Init: given R.sub.0
for k=1 to K
I.sub.k=T(I.sup.IN, I.sup.REF, R.sub.k-1)
G.sub.k=GD(I.sub.k, IREF),
R.sub.k=R.sub.k-1.andgate.G.sub.k (5)
[0064] A "deghosting process" is a process of continuously adding
an aligned region in a ghost region. This is expressed as
follows:
Init: I.sub.k+1.sup.F=I.sup.REF
for k=K to 1
I.sub.k.sup.F=F(I.sub.k, I.sup.F.sub.k+1, G.sub.k) (6)
[0065] Here, the term "F" refers to an ROI-based deghosting
process.
[0066] By performing such iterative calculation, it may be possible
to solve a number of stages of complex motion with a small amount
of calculation.
[0067] FIG. 4 is a block diagram of a cascaded structure of a
plurality of image blenders in FIG. 2 according to an example
embodiment.
[0068] As shown in FIG. 4, the cascaded structure of image blenders
may include a first image blender 100, a second image blender 300,
and a third image blender 500. An overlapped portion between an ROI
detected by an ROI detecting unit 130 of the first image blender
100 and an ROI of an image arranging unit 110 of the first image
blender 100 may be set to an ROI in an image arranging unit 310 of
the second image blender 300. An overlapped portion between an ROI
detected by an ROI detecting unit 330 of the second image blender
300 and an ROI of an image arranging unit 310 of the second image
blender 300 may be set to an ROI in an image arranging unit 510 of
the third image blender 500.
[0069] Further, an image blended by an image blending unit 570 of
the third image blender 500 may be input to the image blending unit
370 of the second image blender 300. An image blended by the image
blending unit 370 of the second image blender 300 may be input to
the image blending unit 170 of the first image blender 100.
[0070] FIG. 5 is a block diagram of an image blender according to
another example embodiment.
[0071] A weight calculating unit 950 may calculate weights using
image information in addition to the ghost region detected as an
ROI to perform natural image blending and narrow a range of ghost
region weights. For HDR image blending, if an image has a great
deal of noise in a ghost region, a decreased range of weights may
greatly reduce the effect of the noise.
[0072] That is, for ROI-based blending:
K i ( x ) = [ 1 - ax i , ax i < 1 0 , otherwise ] ( 7 )
##EQU00004##
[0073] If i=1, the initial ROI may be set as follows:
W ( x ) = { 1 , x .di-elect cons. R 0 , x R ( 8 ) x = ( x 0 , x 1 )
, .alpha. < 1 ( 9 ) ##EQU00005##
[0074] A maximum value of a current weight and a weight decreased
by .alpha. in a horizontal forward direction and a horizontal
backward direction for each line x.sub.0 may be selected. That
is:
[0075] for x1=1 to n
W(x.sub.0, x.sub.1)=max(W(x.sub.0, x.sub.1), W(x.sub.0,
x.sub.1-1)-.alpha.) (10)
[0076] for x1=1 to n
W(x.sub.0, x.sub.1)=max(W(x.sub.0, x.sub.1), W(x.sub.0,
x.sub.1+1)-.alpha.) (11)
[0077] Next, the process may be further performed for x.sub.0 and
x.sub.1 which may be switched. That is, this process may be
performed in a vertical forward direction and a vertical backward
direction column by column. Weighted blending may then be performed
using the weight calculated in this process.
[0078] Further, the weights may be calculated using additional
image information about a region other than the ghost region as an
ROI to perform natural image blending and narrow a range of the
weights. For example, the weights may be set to have a different
attenuation depending on a difference in brightness level which
occurs in the boundary of an ROI. In this case, a may be considered
as:
f(|I.sup.F.sub.k+1(x)-I.sub.K(x)|) (12)
[0079] That is, to reduce a feathering width, a may need to become
larger as a difference in brightness level between two images
blended at the boundary of an ROI becomes smaller.
[0080] For another example, the weights may be set to have a
different attenuation depending on a difference in brightness level
which occurs in the boundary of another ROI. In this case, a may be
considered as:
g(.DELTA.|I.sup.F.sub.k+1|) (13)
[0081] That is, .alpha. may become smaller as an edge value becomes
larger. For an edge portion in an image, the weights may need to be
significantly attenuated to prevent the definition of an image from
deteriorating due to a blurred edge.
[0082] FIG. 6 illustrates an example method for blending images.
The method 600 includes aligning a reference image and an input
image through region of interest (ROI)-based image alignment
(operation 610), detecting an ROI associated with an incompletely
aligned region in the aligned image(operation 620), and blending
the aligned reference image and input image by using a plurality of
weights which are attenuated according to a spatial distance from a
boundary of the detected ROI (operation 630).
[0083] FIG. 7 illustrates another example method for blending
images. The method 700 includes aligning a reference image and an
input image through image alignment based on an n.sup.th region of
interest (ROI) which is detected at a previous stage (operation
710), detecting another ROI associated with an incompletely aligned
region in the aligned image (operation 720), blending the aligned
input image in the other ROI and an (n+1).sup.th blended image,
which is input from a next stage, in a region outside the other ROI
(operation 730), and outputting the last blended image to the
previous stage (operation 740).The processes, functions, methods
and/or software described above may be recorded, stored, or fixed
in one or more computer-readable storage media that includes
program instructions to be implemented by a computer to cause a
processor to execute or perform the program instructions. The media
may also include, alone or in combination with the program
instructions, data files, data structures, and the like. The media
and program instructions may be those specially designed and
constructed, or they may be of the kind well-known and available to
those having skill in the computer software arts. Examples of
computer-readable media include magnetic media, such as hard disks,
floppy disks, and magnetic tape; optical media such as CD-ROM disks
and DVDs; magneto-optical media, such as optical disks; and
hardware devices that are specially configured to store and perform
program instructions, such as read-only memory (ROM), random access
memory (RAM), flash memory, and the like. Examples of program
instructions include machine code, such as produced by a compiler,
and files containing higher level code that may be executed by the
computer using an interpreter. The described hardware devices may
be configured to act as one or more software modules in order to
perform the operations and methods described above, or vice versa.
In addition, a computer-readable storage medium may be distributed
among computer systems connected through a network and
computer-readable codes or program instructions may be stored and
executed in a decentralized manner.
[0084] As a non-exhaustive illustration only, the terminal device
described herein may refer to mobile devices such as a cellular
phone, a personal digital assistant (PDA), a digital camera, a
portable game console, and an MP3 player, a portable/personal
multimedia player (PMP), a handheld e-book, a portable tablet
and/or laptop PC, a global positioning system (GPS) navigation, and
devices such as a desktop PC, a high definition television (HDTV),
an optical disc player, a setup box, and the like consistent with
that disclosed herein.
[0085] A computing system or a computer may include a
microprocessor that is electrically connected with a bus, a user
interface, and a memory controller. It may further include a flash
memory device. The flash memory device may store N-bit data via the
memory controller. The N-bit data is processed or will be processed
by the microprocessor and N may be 1 or an integer greater than 1.
Where the computing system or computer is a mobile apparatus, a
battery may be additionally provided to supply operation voltage of
the computing system or computer.
[0086] It will be apparent to those of ordinary skill in the art
that the computing system or computer may further include an
application chipset, a camera image processor (CIS), a mobile
Dynamic Random Access Memory (DRAM), and the like. The memory
controller and the flash memory device may constitute a solid state
drive/disk (SSD) that uses a non-volatile memory to store data.
[0087] A number of example embodiments have been described above.
Nevertheless, it will be understood that various modifications may
be made. For example, suitable results may be achieved if the
described techniques are performed in a different order and/or if
components in a described system, architecture, device, or circuit
are combined in a different manner and/or replaced or supplemented
by other components or their equivalents. Accordingly, other
implementations are within the scope of the following claims.
* * * * *