METHOD AND DEVICE FOR AVOIDING ROUNDING ERRORS AFTER PERFORMING AN INVERSE DISCRETE ORTHOGONAL TRANSFORMATION

• US 20100208807A1
• Filed: 02/16/2010
• Published: 08/19/2010
• Est. Priority Date: 02/18/2009
• Status: Abandoned Application

• Current Assignee: EcoDisc Technology AG, SONNHALDENSTRASSE 7
• Original Assignee: EcoDisc Technology AG, SONNHALDENSTRASSE 7
• Inventors: SIKORA, Thomas
• Application Number:
  US12/706,499
• Publication Number:
  US 20100208807A1
• Time in Patent Office: Days
• US Class Current: 375/240.120
• CPC Class Codes: G06F 17/147 Discrete orthonormal transforms, e.g. discrete cosine transform, discrete sine transform, and variations therefrom, e.g. modified discrete cosine transform, integer transforms approximating the discrete cosine transform G06F17/145 takes precedence
• CPC Class Codes: H04N 19/45 performing compensation of the inverse transform mismatch, e.g. Inverse Discrete Cosine Transform [IDCT] mismatch
• CPC Class Codes: H04N 19/51 Motion estimation or motion compensation
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/82 involving filtering within a prediction loop
• CPC Class Codes: H04N 19/85 using pre-processing or post-processing specially adapted for video compression

Abstract

The present invention provides a method for avoiding rounding errors during rounding of values after performing an inverse discrete orthogonal transformation. In a first step, a block of coefficients (F′[u][v]) is transformed into a block of image pixel values (f′[y][x]) by means of an inverse discrete orthogonal transformation, wherein each image pixel comprises an image pixel value. In a second step, a product of the block of image pixel values (f′[y][x]) with a first pixel pattern (A[y][x]) is calculated. In a third step, the product values of the second step are summed up to a first sum (Sum1′). In a fourth step, it is determined whether the first sum (Sum1′) is even or odd. In a fifth step, if the first sum (Sum1′) is even, a block of manipulation values (C[y][x]) is added to or subtracted from the block of image pixel values (f′[y][x]) for generating a manipulated block of image pixel values (f[y][x]), wherein the block of manipulation values (C[y][x]) is formed in such a way that rounding errors are avoided in a subsequent rounding operation.

Claims

1. Method for avoiding rounding errors during rounding of values after performing an inverse discrete orthogonal transformation (7, 23), wherein the method comprises the following steps in the following order:

a) Transforming a block of coefficients (F′

[u][v]) into a block of image pixel values (f′

[y][x]) by means of an inverse discrete orthogonal transformation (7, 23), wherein each image pixel comprises a image pixel value (S1).

b) Calculating a product of the block of image pixel values (f′

[y][x]) with a first pixel pattern (A[y][x]) (S2);

c) summing up the product values of step b) (S3) to a first sum (Sum1′

);

d) determining whether the first sum (Sum1′

) is even or odd (S4); and

e) if the first sum (Sum1′

) is even, adding or subtracting a block of manipulation values (C[y][x]) to the block of image pixel values (f′

[y][x]) for generating a manipulated block of image pixel values (f[y][x]), wherein the block of manipulation values (C[y][x]) is formed in such a way that rounding errors in a subsequent rounding operation are minimized (S11, S12, S13).

2. Method according to claim 1, wherein in step (e) the following steps are performed, if the first sum (Sum1′

) is even;

e1) Calculating a product of the block of image pixel values (f′

[y][x]) with a second pixel pattern (B[y][x]) (S7);

e2) summing up the product values from step e1) (S8) to a second sum (Sum2′

);

wherein

e3) if the second sum (Sum2′

) is even (S9, S10), the block of manipulation values (C[y][x]) is added to the block of image pixel values (f′

[y][x]) (S11); and

e4) if the second sum (Sum2′

) is odd (S9, S10), the block of manipulation values (C[y][x]) is subtracted from the block of image pixel values (S12).

3. Method according to claim 1, wherein the first pixel pattern (A[y][x]), the second pixel pattern (B[y][x]) or the block of manipulation values (C[y][x]), or any combination thereof is a numerically or analytically optimized matrix.

4. Method according to claim 3, wherein the matrix of the first pixel pattern (A[y][x]), the matrix of the second pixel pattern (B [y][x]) or the matrix of the block of manipulation values (C[y][x]), or any combination thereof, is numerically optimized by the gradient descent method;

or stochastic programming;

or analytically optimized by the Wiener filter method, or any combination thereof.

5. Method according to claim 1, wherein the matrix of the first pixel pattern (A[y][x]) is the following matrix:

6. Method according to claim 2, wherein the second pixel pattern (B[y][x]) consists of a quadratic symmetric matrix.

7. Method according to claim 1, wherein the block of manipulation values (C[y][x]) consists of a quadratic symmetric matrix.

8. Method according to claim 1, wherein the method further comprises the following step:

f) Rounding each manipulated image pixel value of the manipulated block of image pixel values (f[y][x]) to an integer number (S13).

9. Method for providing a compressed video signal comprising the following steps:

a) Performing a prediction coding by means of a comparison of subsequent images;

b) transforming image blocks into blocks of transformation coefficients by means of a discrete cosine transformation;

c) performing the method for avoiding rounding errors according to claim 1 in an inverse branch of an encoder in which an encoded image is decoded.

10. A storage medium in which the compressed video signal according to claim 9 is stored.

11. Device for avoiding rounding errors (30) during rounding of values after performing an inverse discrete orthogonal transformation (7, 23), wherein the device comprises means for performing the method according to claim 1.

12. Decoder (1), in which the method according to claim 1 is performed for avoiding rounding errors.

13. Encoder (20) in which the method according to claim 1 is performed for avoiding rounding errors.

14. Computer program comprising instructions which performs the steps of the method according to claim 1 when running on a computer.

15. Method according to claim 2, wherein the second pixel pattern (B[y][x]) consists of the following matrix:

16. Method according to claim 1, wherein the block of manipulation values (C[y][x]) consists of the following matrix: