Method and apparatus for encoding and decoding multi-view video using image stitching

1. A method of encoding multi-view video data, comprising the steps of:

decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

encoding said stitched image using a first encoding algorithm;

encoding the overlapped region of each view image using a second encoding algorithm,

wherein said decomposing step includes the steps of:

estimating disparity information for each view image, based on a predetermined view image; and

decomposing each view image into said overlapped region and said non-overlapped region using the estimated disparity information, and

wherein said step of estimating the disparity information includes estimating a global disparity vector based on an adjacent view image toward the middle view direction, and

wherein said step of estimating the global disparity vector comprises the steps of:

reducing the size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

estimating an initial global disparity value for each of the reduced view images; and

estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate.

2. The method according to claim 1 , wherein the stitched image is generated by stitching the non-overlapped regions of each view image to a boundary of the middle view image.

3. The method according to claim 1 , wherein the first encoding algorithm is a motion-compensated encoding algorithm.

4. The method according to claim 1 , wherein the second encoding algorithm is a motion- and disparity-compensated encoding algorithm.

5. A method of encoding multi-view video data, comprising the steps of:

decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

encoding said stitched image using a first encoding algorithm;

encoding the overlapped region of each view image using a second encoding algorithm,

wherein said decomposing step includes the steps of:

estimating disparity information for each view image, based on a predetermined view image; and

decomposing each view image into said overlapped region and said non-overlapped region using the estimated disparity information, and

wherein said step of estimating the disparity information includes estimating a global disparity vector based on the middle view image, and

wherein said step of estimating the global disparity vector comprises the steps of:

reducing the size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

estimating an initial global disparity value for each of the reduced view images; and

estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate.

6. The method according to claim 1 or 5 , wherein the predetermined size of the block unit is selected from 4×4 and 8×8 blocks.

7. The method according to claim 1 , wherein said step of encoding the stitched image using the first encoding algorithm comprises the steps of:

partitioning the middle view image into a plurality of slices in a desired size and encoding each of the plurality of the slices, on a slice basis; and

encoding the non-overlapped regions of each view image, on a view basis.

8. The method according to claim 7 , wherein a slice address of the overlapped region of each view image is included in a header of the encoded middle view image streams.

9. A method of decoding multi-view video data, comprising the steps of:

decoding encoded streams of a stitched image including a middle view image and a non-overlapped region of each view image of the multi-view video using a motion compensated decoding algorithm;

decoding encoded streams of an overlapped region of each view image using a motion- and disparity-compensated decoding algorithm, based on the decoded stitched image, and

wherein said disparity-compensated decoding algorithm decodes an encoded multi-view video data, comprising the steps of:

decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

encoding said stitched image using a first encoding algorithm;

encoding the overlapped region of each view image using a second encoding algorithm,

wherein said decomposing step includes the steps of:

estimating disparity information for each view image, based on a predetermined view image; and

decomposing each view image into said overlapped region and said non-overlapped region using the estimated disparity information, and

wherein said step of estimating the disparity information includes estimating a global disparity vector based on an adjacent view image toward the middle view direction, and

wherein said step of estimating the global disparity vector comprises the steps of:

reducing the size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

estimating an initial global disparity value for each of the reduced view images; and

estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate; and

restoring the multi-view video by combining the decoded non-overlapped and overlapped regions of each view image.

10. The method according to claim 9 , further comprising a step of restoring a specific view image of the multi-view video in response to a user' selection, wherein said step of restoring the specific view image includes:

decoding encoded streams for the non-overlapped region of the specific view image; and

decoding the encoded streams for the slice(s) of the middle view image overlapped with the specific view image, by using the slice address that is included in the header of the encoded streams of the middle view image.

11. An apparatus for encoding multi-view video data, comprising:

region decomposing means for decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

image stitching means for generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

a first encoding means for encoding said stitched image using a first encoding algorithm;

a second encoding means for encoding the overlapped region of each view image using a second encoding algorithm; and

a disparity estimation means for estimating disparity information for each view image, based on a predetermined view image,

wherein said region decomposing means decomposes each view image into said overlapped region and said non-overlapped region using the estimated disparity information,

wherein said disparity estimation means estimates a global disparity vector based on an adjacent view image toward the middle view direction, and

wherein said disparity estimation means comprises:

sub-sampling means for reducing a size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

initial disparity estimation means for estimating an initial global disparity value of each reduced view images; and

boundary region-based disparity estimation means for estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate.

12. The apparatus according to claim 11 , wherein said image stitching means generates said stitched image by stitching the non-overlapped regions of each view image to a boundary of the middle view image.

13. An apparatus for encoding multi-view video data, comprising:

region decomposing means for decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

image stitching means for generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

a first encoding means for encoding said stitched image using a first encoding algorithm;

a second encoding means for encoding the overlapped region of each view image using a second encoding algorithm; and

a disparity estimation means for estimating disparity information for each view image, based on a predetermined view image, wherein said region decomposing means decomposes each view image into said overlapped region and said non-overlapped region using the estimated disparity information,

wherein said disparity estimation means estimates a global disparity vector based on the middle view image, and

wherein said disparity estimation means comprises:

sub-sampling means for reducing a size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

initial disparity estimation means for estimating an initial global disparity value of each reduced view images; and

boundary region-based disparity estimation means for estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate.

14. The apparatus according to claim 11 , wherein the first encoding means comprises:

means for partitioning the middle view image into a plurality of slices in a desired size and encoding each of the plurality of the slices, on a slice basis; and

means for encoding the non-overlapped regions of each view image, on a view basis.

15. The apparatus according to claim 14 , wherein a slice address of the overlapped region of each view image is included in a header of the encoded middle view image streams.

16. An apparatus for decoding multi-view video data, comprising:

means for receiving encoded streams of an overlapped region of each view image of the multi-view video and encoded streams of a stitched image including a middle view image and a non-overlapped region of each view image of the multi-view video;

a first decoding means for decoding the encoded streams of the stitched image using a motion compensated decoding algorithm;

a second decoding means for decoding the encoded streams of the overlapped region of each view image using motion- and disparity-compensated decoding algorithm, based on the stitched image, and

wherein the disparity-compensated decoding algorithm decodes an encoding multi-view video data generated from an apparatus, comprising:

region decomposing means for decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

image stitching means for generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

a first encoding means for encoding said stitched image using a first encoding algorithm;

a second encoding means for encoding the overlapped region of each view image using a second encoding algorithm; and

a disparity estimation means for estimating disparity information for each view image, based on a predetermined view image,

wherein said region decomposing means decomposes each view image into said overlapped region and said non-overlapped region using the estimated disparity information,

wherein said disparity estimation means estimates a global disparity vector based on an adjacent view image toward the middle view direction, and

wherein said disparity estimation means comprises:

sub-sampling means for reducing a size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

initial disparity estimation means for estimating an initial global disparity value of each reduced view images; and

boundary region-based disparity estimation means for estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate; and

restoring means for restoring the multi-view video by combining the decoded non-overlapped and overlapped regions of each view image.

17. The apparatus according to claim 16 , wherein said apparatus is capable of restoring a specific view image of the multi-view video in response to a user' selection by decoding encoded streams for the non-overlapped region of the specific view image; decoding the encoded streams for the slice(s) of the middle view image overlapped with the specific view image, using the slice address that is included in the header of the encoded streams of the middle view image; and combining the decoded non-overlapped region and the decoded slice(s) of the middle view image overlapped with the specific view image.

18. A method of encoding multi-view video data, comprising the steps of:

decomposing each view image of the multi-view video into an overlapped region and a non-overlapped region, said overlapped region being overlapped with other view image and said non-overlapped region not being overlapped with other view image;

generating a stitched image by combining said non-overlapped region of each view image and a middle view image;

encoding said stitched image using a first encoding algorithm;

encoding the overlapped region of each view image using a second encoding algorithm,

wherein said decomposing step includes the steps of:

estimating disparity information for each view image, based on a predetermined view image,

wherein said step of estimating the global disparity vector further comprises the steps of:

reducing the size of each view image by sub-sampling one pixel from each block unit of each view image, said block unit having a predetermined size;

estimating an initial global disparity value for each of the reduced view images; and

estimating the global disparity vector for a boundary region of each view image, said boundary region being defined by multiplying said estimated initial global disparity value by the sub-sampling rate; and

decomposing each view image into said overlapped region and said non-overlapped region using the estimated disparity information, and

wherein said step of estimating the disparity information includes estimating a global disparity vector based on the middle view image.