VIDEO ENCODING/DECODING APPARATUS AND ADAPTIVE OVERLAPPED BLOCK MOTION COMPENSATION METHOD AND APPARATUS EMPLOYING ADAPTIVE WEIGHTS THEREFOR

• US 20110200098A1
• Filed: 10/14/2009
• Published: 08/18/2011
• Est. Priority Date: 10/21/2008
• Status: Active Grant

• Current Assignee: SK Telecom Co., Ltd.
• Original Assignee: SK Telecom Co., Ltd.
• Inventors: Choe, Yoonsik, Kim, Hayoon, Choi, Yungho, Kim, Yonggoo
• Granted Patent: US 8,503,522 B2
• Time in Patent Office: Days
• US Class Current: 375/240.02
• CPC Class Codes: H04N 19/567 Motion estimation based on rate distortion criteria
• CPC Class Codes: H04N 19/583 Motion compensation with overlapping blocks

Abstract

The present disclosure provides an adaptive overlapped block motion compensation apparatus employing adaptive weights including: an R_BMC(x,y) calculator for calculating residual signal R_BMC(x,y) in block motion estimation; a D_N(x,y) calculator for calculating differential pixel value D_N(x,y) between a motion compensated block generated by a motion vector of an adjacent block and a currently motion compensated block; a unit for updating and setting additional information for performing an adaptive overlapped block motion compensation based on R_BMC(x,y) and D_N(x,y); and an adaptive overlapped block motion compensation operating on respective pixels in certain pixel units based on the additional information, progressing to subsequent pixels if an absolute value of R_BMC(x,y) is not greater than the additional information, and if the absolute value of R_BMC(x,y) is greater than the additional information, calculating an R_OBmc(x,y) with an H_N(x,y) selectively signed so that an absolute value of R_OBMC(_x,y) is smaller than the absolute value of R_BMC(x,y).

Claims

1. An adaptive overlapped block motion compensation apparatus employing adaptive weights comprising:

an R_BMC(x,y) calculator for performing a calculation of a residual signal R_BMC(x,y) in a block motion estimation;

a D_N(x,y) calculator for performing a calculation of a differential pixel value D_N(x,y) between a motion compensated block generated by a motion vector of an adjacent block and a currently motion compensated block through a currently estimated motion vector;

an additional information updating/setting unit for updating and setting additional information for performing an adaptive overlapped block motion compensation based on the R_BMC(x,y) and the D_N(x,y); and

an adaptive overlapped block motion compensator for performing the adaptive overlapped block motion compensation for respective pixels in certain pixel units based on the additional information, progressing the adaptive overlapped block motion compensation over to subsequent pixels as long as an absolute value of the R_BMC(x,y) is smaller than or equal to the additional information, and if the absolute value of the R_BMC(x,y) is greater than the additional information, calculating a residual signal R_OBMC(_x,y) with an H_N(x,y) selectively signed so that an absolute value of the R_OBMC(x,y) in an overlapped block motion reconstruction is smaller than the absolute value of the R_BMC(x,y).

2. The adaptive overlapped block motion compensation apparatus employing adaptive weights of claim 1, wherein the additional information updating/setting unit comprising:

an H_N(x,y) sign determiner for making a determination on a sign of the H_N(x,y) that is a weight based on a sign of the R_BMC(x,y) after the calculation and a sign of the D_N(x,y) after the calculation;

an R_OBMC(_x,y) calculator for performing a calculation of the residual signal R_OBMC(x,y) in an overlapped block motion reconstruction based on the sign of the H_N(x,y) upon the determination;

a reference pixel updater for updating a first reference pixel or a second reference pixel on the basis of the sign of the R_BMC(x,y) after the calculation and a sign of the R_OBMC(x,y) after the calculation so that the first reference pixel is updated to be smaller one of an absolute value of the first reference pixel preceding the R_OBMC(x,y) and an absolute value of the R_OBMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are same, and the second reference pixel is updated to be larger one of an absolute value of the second reference pixel preceding the R_BMC(x,y) and the absolute value of the R_BMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_ouvic(x,y) are not same; and

an additional information setter for setting larger one of the first reference pixel and the second reference pixel as the additional information.

3. The adaptive overlapped block motion compensation apparatus employing adaptive weights of claim 2, wherein the H_N(x,y) sign determiner in a motion estimation determines the sign of the H_N(x,y) so that the R_BMC(x,y) valued positive causes H_N(x,y) x D_N(x,y) to be signed negative and determines the sign of the H_N(x,y) so that the R_BMC(x,y) valued negative causes H_N(x,y) x D_N(x,y) to be signed positive.

4. The adaptive overlapped block motion compensation apparatus employing adaptive weights of claim 2, wherein the R_OBMC(x,y) calculator and the adaptive overlapped block motion compensator calculate the R_OBMC(x,y) based on a formula of R_OBMC(x,y)=R_BMC(x,y)+H_N(x,y)×

D_N(x,y).

5. The adaptive overlapped block motion compensation apparatus employing adaptive weights of claim 2, wherein the reference pixel updater performs the updating of the first reference pixel and the second reference pixel sequentially with respect to all pixels in the certain pixel units.

6. The adaptive overlapped block motion compensation apparatus employing adaptive weights of claim 1, wherein the adaptive overlapped block motion compensator determines a sign of the H_N(x,y) so that the R_BMC(x,y) and H_N(x,y) x D_N(x,y) are signed opposite to each other.

7. An adaptive overlapped block motion compensation method employing adaptive weights comprising:

performing a calculation of a residual signal R_BMC(x,y) in a block motion estimation;

performing a calculation of a differential pixel value D_N(x,y) between a motion compensated block generated by a motion vector of an adjacent block and a currently motion compensated block through a currently estimated motion vector;

updating/setting additional information for performing an adaptive overlapped block motion compensation based on the R_BMC(x,y) and the D_N(x,y); and

performing the adaptive overlapped block motion compensation for respective pixels in certain pixel units based on the additional information, progressing the adaptive overlapped block motion compensation over to subsequent pixels as long as an absolute value of the R_BMC(x,y) is smaller than or equal to the additional information, and if the absolute value of the R_BMC(x,y) is greater than the additional information, calculating a residual signal R_OBMC(x,y) with an H_N(x,y) selectively signed so that an absolute value of the R_OBMC(x,y) in an overlapped block motion reconstruction is smaller than the absolute value of the R_BMC(x,y).

8. The adaptive overlapped block motion compensation method employing adaptive weights of claim 7, wherein the step of updating/setting additional information comprising:

performing a determination on a sign of the H_N(x,y) that is a weight based on a sign of the R_BMC(x,y) after the calculation and a sign of the D_N(x,y) after the calculation;

performing a calculation of the residual signal R_OBMC(_x,y) in an overlapped block motion reconstruction based on the sign of the H_N(x,y) upon the determination;

updating a reference pixel of a first reference pixel or a second reference pixel based on the sign of the R_BMC(x,y) after the calculation and a sign of the R_OBMC(x,y) after the calculation so that the first reference pixel is updated to be smaller one of an absolute value of the first reference pixel preceding the R_OBMC(x,y) and an absolute value of the R_OBMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are same, and the second reference pixel is updated to be larger one of an absolute value of the second reference pixel preceding the R_BMC(x,y) and an absolute value of the R_BMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are not same; and

setting larger one of the first reference pixel and the second reference pixel as the additional information.

9. The adaptive overlapped block motion compensation method employing adaptive weights of claim 8, wherein the step of performing the determination of the sign of the H_N(x,y) determines the sign of the H_N(x,y) so that the R_BMC(x,y) valued positive causes H_N(x,y) x D_N(x,y) to be signed negative and determines the sign of the H_N(x,y) so that the R_BMC(x,y) valued negative causes H_N(x,y)×

D_N(x,y) to be signed positive.

10. The adaptive overlapped block motion compensation method by variable units of claim 9, wherein the constant k is determined in consideration of a ratio of compression efficiency to a computation cost.

11. The adaptive overlapped block motion compensation method by variable units of claim 8, wherein the step of updating the reference pixel performs the updating of the first reference pixel and the second reference pixel sequentially with respect to all pixels in the certain pixel units.

12. The adaptive overlapped block motion compensation method by variable units of claim 7, wherein the step of performing the adaptive overlapped block motion compensation determines a sign of the H_N(x,y) so that the R_BMC(x,y) and H_N(x,y)×

D_N(x,y) are signed opposite to each other.

13. A video encoding apparatus comprising:

a motion estimator/compensator for performing an adaptive overlapped block motion compensation for respective pixels in certain pixel units based on additional information having been set, progressing the adaptive overlapped block motion compensation over to subsequent pixels as long as an absolute value of a residual signal R_BMC(x,y) in a block motion reconstruction is smaller than or equal to the set additional information, and if the absolute value of the R_BMC(x,y) is greater than the additional information, calculating a residual signal R_OBMC(x,y) with an H_N(x,y) selectively signed so that an absolute value of the R_OBMC(X,Y) in an overlapped block motion reconstruction is smaller than the absolute value of the R_BMC(x,y) to predict predicted pixel values of respective pixels in an image;

a subtractor for calculating a difference between an original pixel value of each pixel of a current block and a predicted pixel value of each pixel of the current block to generate a residual signal;

a transformer for performing a transform of the residual signal into frequency coefficients;

a quantizer for performing a quantization of the frequency coefficients after the transform; and

an encoder for encoding the frequency coefficients after the quantization into a bitstream.

14. The video encoding apparatus of claim 13, wherein the motion estimator/compensator calculates the R_OBMC(x,y) based on a formula of R_OBMC(x,y)=R_BMC(x,y)+H_N(x,y)×

D_N(x,y).

15. The video encoding apparatus of claim 14, wherein the motion estimator/compensator determines a sign of the H_N(x,y) so that the R_BMC(x,y) and H_N(x,y)×

D_N(x,y) are signed opposite to each other.

16. The video encoding apparatus of claim 13, wherein the motion estimator/compensator makes a determination on a sign of the H_N(x,y) based on a sign of the R_BMC(x,y) and a differential pixel value D_N(x,y) between a motion compensated block generated by a motion vector of an adjacent block and a currently motion compensated block;

performs a calculation of a residual signal R_OBMC(x,y) in an overlapped block motion reconstruction based on the sign of the H_N(x,y) upon the determination;

updates a first reference pixel or a second reference pixel based on a sign of the R_OBMC(x,y) after the calculation and the sign of the R_BMC(x,y) so that the first reference pixel is updated to be smaller one of an absolute value of the first reference pixel preceding the R_OBMC(x,y) and an absolute value of the R_OBMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are same, and the second reference pixel is updated to be larger one of an absolute value of the second reference pixel preceding the R_BMC(x,y) and the absolute value of the R_BMC(x,y) if the sign of the _RBMc(x,y) and the sign of the R_OBMC(x,y) are not same; and

sets larger one of the first reference pixel and the second reference pixel as the additional information.

17. A video decoding apparatus comprising:

a comparator for comparing an absolute value of a residual signal R_BMCT(x,y) in a block motion reconstruction to additional information, the R_BMCT(x,y) having been encoded and transmitted and the additional information having been set; and

an adaptive overlapped block motion compensator for performing an adaptive overlapped block motion compensation on each of successive pixels as long as the absolute value of the R_BMCT(x,y) is smaller than or equal to the additional information, and if the absolute value of the R_BMCT(x,y) is greater than the additional information, calculating an R_OBMC(x,y) with an H_N(x,y) selectively signed so that an absolute value of the R_OBMC(x,y) is greater than the absolute value of the R_BMCT(x,y).

18. The video decoding apparatus of claim 17, wherein the adaptive overlapped block motion compensator calculates the R_OBMC(x,y) based on a formula of R_OBMC(x,y)=R_BMCT(x,y)−

H_N(x,y)×

D_N(x,y).

19. The video decoding apparatus of claim 18, wherein the adaptive overlapped block motion compensator determines a sign of the H_N(x,y) so that the R_BMCT(x,y) and H_N(x,y)×

D_N(x,y) are signed identical to each other.

20. The video decoding apparatus of claim 17, wherein the additional information is generated and provided by an encoding apparatus and the encoding apparatus makes a determination on a sign of the H_N(x,y) that is a weight based on a sign of an R_BMC(x,y) and a differential pixel value D_N(x,y) between a motion compensated block generated by a motion vector of an adjacent block and a currently motion compensated block;

performs a calculation of a residual signal R_OBMC(x,y) in an overlapped block motion reconstruction based on the sign of the H_N(x,y) upon the determination;

performs an update of a first reference pixel or a second reference pixel based on a sign of the R_OBMC(x,y) after the calculation and the sign of the R_BMC(x,y) so that the first reference pixel is updated to be smaller one of an absolute value of the first reference pixel preceding the R_OBMC(x,y) and an absolute value of the R_OBMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are same, and the second reference pixel is updated to be larger one of an absolute value of the second reference pixel preceding the R_BMC(x,y) and the absolute value of the R_BMC(x,y) if the sign of the R_BMC(x,y) and the sign of the R_OBMC(x,y) are not same; and

upon completing the update of every pixel in certain pixel units, sets larger one of the first reference pixel and the second reference pixel as the additional information.