Method and apparatus for correcting errors in a transmitted video signal

• US 5,751,361 A
• Filed: 12/20/1996
• Issued: 05/12/1998
• Est. Priority Date: 12/23/1995
• Status: Expired due to Term

• Current Assignee: Wi-LAN Inc.
• Original Assignee: Daewoo Electronics Company Limited (Dayou Holdings Co.,Ltd.)
• Inventors: Kim, Jin-Hun
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US08/769,831
• Time in Patent Office: 508 Days
• Field of Search: 348/384, 348/409, 348/420, 348/469, 348/607, 348/616, 386/50
• US Class Current: 375/240.12
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/65 using error resilience
• CPC Class Codes: H04N 19/895 in combination with error concealment

Abstract

An apparatus for use in a video signal decoding system for concealing errors in a video signal first detects a lost block in the video signal, the lost block having errors, and then provides pixel values of neighboring blocks for the lost block in the video signal based on information for the lost block. Subsequently, The apparatus computes edge gradients of neighboring pixels included in a predetermined range of pixels surrounding the lost block based on the pixel values of the neighboring blocks and determines a pixel interpolation direction based on the edge gradients of neighboring pixels. A spatially interpolated block is produced based on the pixel values of the neighboring blocks and the pixel interpolation direction and it is provided as a substitution block for compensating the lost block.

Claims

1. A method, for use in a video signal decoding system, for concealing errors in a video signal transmitted in a compressed form, wherein the video signal is divided into a plurality of blocks each of which has N×

M pixel values with N and M being positive integers, the method comprising the steps of;

(a) detecting a lost block containing errors;

(b) providing pixel values of blocks neighboring the lost block in the video signal;

(c) calculating an edge gradient at each of neighboring pixels included in a predetermined range of pixels surrounding the lost block based on the pixel values of the neighboring blocks and computing a magnitude and an angle of each of the edge gradients;

(d) generating a spatially interpolated block based on the edge gradients and the pixel values of the neighboring blocks; and

(e) providing the spatially interpolated block as a substitution block for compensating the lost block,

wherein the step (d) includes the steps of;

(d1) classifying the angle of the edge gradient at each of the neighboring pixels along with one of predetermined directional indicators with which the angle of the edge gradient is most closely aligned;

(d2) if a line drawn through a neighboring pixel location along the direction determined by the angle of the gradient at the neighboring pixel passes through the lost block, determining that particular pixel as a voting pixel;

(d3) summing up the magnitudes of the gradients aligned with each directional indicator in order to mutually exclusively accumulate the directional indicators of the voting pixels;

(d4) selecting the directional indicator exhibiting a largest accumulated sum as a pixel interpolation direction by comparing the sum of the magnitudes for each of the directional indicators with one another;

(d5) comparing the largest accumulated sum with a predetermined threshold value; and

(d6) if the largest accumulated sum is larger than or equal to the predetermined threshold value, producing the spatially interpolated block through the use of a directional interpolation based on pixel values of the neighboring blocks along the pixel interpolation direction.

2. The method as recited in claim 1, wherein the edge gradients are computed through the use of a Sobel operator.

3. The method as recited in claim 1, wherein the edge gradients are determined for three or four rows and columns of pixels surrounding the lost block.

4. The method as recited in claim 1, wherein the directional indicators contain four indicators, representing horizontal, vertical, diagonally ascending and diagonally descending directions, respectively.

5. The method as recited in claim 1, wherein the step (d6) further includes the steps of:

(d611) computing an average value of all pixel values of each of the neighboring blocks along the pixel interpolation direction;

(d612) calculating a mean value of the average values for the neighboring blocks along the pixel interpolation direction; and

(d613) generating the spatially interpolated block whose pixel values are determined as the mean value.

6. The method as recited in claim 1, wherein the step (d6) further includes the step of:

(d631) generating, if the largest accumulated sum is smaller than the predetermined threshold value, the spatially interpolated block based on a weighted mean value of all pixel values of the neighboring blocks for the lost block.

7. The method as recited in claim 1, wherein the step (d6) further includes the steps of:

(d621) converting the neighboring blocks along the pixel interpolation direction to transformed neighboring blocks each of which has N×

M number of DCT coefficients through the use of a discrete cosine transform;

(d622) generating a directionally interpolated block through the use of a directional interpolation based on the transformed neighboring blocks along the pixel interpolation direction; and

(d623) producing the spatially interpolated block by inversely transforming the directionally interpolated block through the use of an inverse discrete cosine transform.

8. The method as recited in claim 7, wherein each DCT coefficient of the directionally interpolated block is determined by averaging DCT coefficients selected from the transformed neighboring blocks along the pixel interpolation direction, said selected DCT coefficients being of a same frequency in the transformed neighboring blocks along the pixel interpolation direction.

9. The method as recited in claim 1, wherein the step (d6) further includes the steps of:

(d641) converting the neighboring blocks for the lost block to transformed neighboring blocks each of which has N×

M number of DCT coefficients through the use of a discrete cosine transform;

(d642) producing, if the largest accumulated sum is smaller than the predetermined threshold value, an averaged block based on the transformed neighboring blocks for the lost block; and

(d643) producing the spatially interpolated block by inversely transforming the averaged block through the use of an inverse discrete cosine transform.

10. The method as recited in claim 9, wherein each DCT coefficient S_ij of the averaged block is determined as:

##EQU5## wherein W_k represents a predetermined weight factor for a kth transformed neighboring block for the lost block;

Bk_ij denotes an ijth DCT coefficient of the kth transformed neighboring block;

k ranges from 1 to 8;

i ranges from 1 to N; and

j ranges from 1 to M.

11. An apparatus, for use in a video signal decoding system, for concealing errors in a video signal transmitted in a compressed form, wherein the video signal is divided into a plurality of blocks each of which has N×

M pixel values with N and M being positive integers, which comprises;

means for detecting a lost block containing errors;

means for providing pixel values of blocks neighboring the lost block in the video signal;

a gradient calculation means for computing an edge gradient at each of neighboring pixels included in a predetermined range of pixels surrounding the lost block based on the pixel values of the neighboring blocks and calculating a magnitude and an angle of each of the edge gradients;

a direction determination means for determining a pixel interpolation direction based on the edge gradients of neighboring pixels;

a block generation means for producing a spatially interpolated block based on the pixel values of the neighboring blocks and the pixel interpolation direction; and

a contribution means for providing the spatially interpolated block as a substitution block for compensating the lost block,

wherein the gradients are computed through the use of a Sobel operator.

12. The apparatus according to claim 11, wherein the gradient calculation means ciphers the edge gradients of three or four rows and columns of pixels surrounding the lost block.

13. An apparatus, for use in a video signal decoding system, for concealing errors in a video signal transmitted in a compressed form, wherein the video signal is divided into a plurality of blocks each of which has N×

M pixel values with N and M being positive integers, which comprises;

means for detecting a lost block containing errors;

means for providing pixel values of blocks neighboring the lost block in the video signal;

a gradient calculation means for computing an edge gradient at each of neighboring pixels included in a predetermined range of pixels surrounding the lost block based on the pixel values of the neighboring blocks and calculating a magnitude and an angle of each of the edge gradients;

a direction determination means for determining a pixel interpolation direction based on the edge gradients of neighboring pixels;

a block generation means for producing a spatially interpolated block based on the pixel values of the neighboring blocks and the pixel interpolation direction; and

a contribution means for providing the spatially interpolated block as a substitution block for compensating the lost block,

wherein the direction determination means includes;

means for classifying the angle of the edge gradient at each of the neighboring pixels along with one of predetermined directional indicators with which the angle of the edge gradient is most closely aligned;

means for determining, if a line drawn through a neighboring pixel location along the direction determined by the angle of the gradient at the neighboring pixel passes through the lost block, that particular pixel as a voting pixel;

means for summing up the magnitudes of the gradients aligned with each directional indicator in order to mutually exclusively accumulate the directional indicators of the voting pixels; and

means for selecting the directional indicator exhibiting a largest accumulated sum as the pixel interpolation direction by comparing the sum of the magnitudes for each of the directional indicators with one another.

14. The apparatus according to claim 13, wherein the directional indicators contain four indicators, representing horizontal, vertical, diagonally ascending and diagonally descending directions, respectively.

15. The apparatus according to claim 13, wherein the block generation means produces the spatially interpolated block through the use of a directional interpolation based on the neighboring blocks corresponding to the pixel interpolation direction.

16. The apparatus according to claim 15, wherein the block generation means includes:

means for computing an average value of all pixel values of each of the neighboring blocks along the pixel interpolation direction;

means for calculating a mean value of the average values for the neighboring blocks along the pixel interpolation direction; and

means for generating the spatially interpolated block whose pixel values are determined as the mean value.

17. The apparatus according to claim 13, wherein the block generation means includes:

means for converting the neighboring blocks for the lost block to transformed neighboring blocks for the lost block each of which has N×

M number of DCT coefficients through the use of a discrete cosine transform;

means for generating a directionally interpolated block through the use of a directional interpolation based on the transformed neighboring blocks along the pixel interpolation direction; and

means for producing the spatially interpolated block by inversely transforming the directionally interpolated block through the use of an inverse discrete cosine transform.

18. The apparatus according to claim 17, wherein each DCT coefficient of the directionally interpolated block is determined by averaging DCT coefficients selected from the transformed neighboring blocks along the pixel interpolation direction, said selected DCT coefficients being of a same frequency in the transformed neighboring blocks along the pixel interpolation direction.

19. The apparatus according to claim 13, wherein the block generation means includes:

means for generating a directionally interpolated block through the use of a directional interpolation based on the neighboring blocks along the pixel interpolation direction;

means for producing an averaged block whose pixel values are determined as a weighted mean value of pixel values of the neighboring blocks for the lost block;

a comparison means for comparing the largest accumulated sum with a predetermined threshold value to thereby provide comparison signals; and

a selection means for choosing, in response to the comparison signals, either the directionally interpolated block or the averaged block as the spatially interpolated block.

20. The apparatus according to claim 19, wherein the comparison means provides a first comparison signal if the largest accumulated sum is larger than or equal to the predetermined threshold value and, if otherwise, supplies a second comparison signal.

21. The apparatus according to claim 20, wherein the selection means includes:

means for providing, in response to the first comparison signal, the directionally interpolated block as the spatially interpolated block; and

means for supplying, in response to the second comparison signal, the averaged block as the spatially interpolated block.

22. The apparatus according to claim 13, wherein the block generation means includes:

means for converting the neighboring blocks for the lost block to transformed neighboring blocks for the lost block each of which has N×

M number of DCT coefficients through the use of a discrete cosine transform;

means for generating a directionally interpolated block through the use of a directional interpolation based on the transformed neighboring blocks along the pixel interpolation direction;

means for producing an averaged block based on the transformed neighboring blocks for the lost block;

a comparison means for comparing the largest accumulated sum with a predetermined threshold value to thereby provide comparison signals;

a selection means for choosing, in response to the comparison signals, either the directionally interpolated block or the averaged block; and

means for producing the spatially interpolated block by inversely transforming the selected block through the use of an inverse discrete cosine transform.

23. The apparatus according to claim 22, wherein each DCT coefficient S_ij of the averaged block is determined as:

##EQU6## wherein W_k represents a predetermined weight factor for a kth transformed neighboring block for the lost block;

Bk_ij denotes an ijth DCT coefficient of the kth transformed neighboring block;

k ranges from 1 to 8;

i ranges from 1 to N; and

j ranges from 1 to M.

24. The apparatus according to claim 23, wherein the comparison means provides a first comparison signal if the largest accumulated sum is larger than or equal to the predetermined threshold value and, if otherwise, supplies a second comparison signal.

25. The apparatus according to claim 24, wherein the selection means includes:

means for providing, in response to the first comparison signal, the directionally interpolated block as the selected block; and

means for supplying, in response to the second comparison signal, the averaged block as the selected block.