Video block transform

• US 20030093452A1
• Filed: 08/22/2002
• Published: 05/15/2003
• Est. Priority Date: 08/23/2001
• Status: Active Grant

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Zhou, Minhua
• Granted Patent: US 7,185,037 B2
• Time in Patent Office: Days
• US Class Current: 708/520
• 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 19/61 in combination with predictive coding

Abstract

Integer transforms of 4×4 blocks of the type used in proposed H.26L but adapted to 16-bit arithmetic by shifts after matrix multiplications and integer approximation matrices absorbing scaling factors.

Claims

1. A method of block transformation, comprising:

(a) providing an input 4×

4 block Y of integer coefficients; and

(b) computing an output 4×

4 block as N Y N^T where M is the 4×

4 matrix

$N=\left[\begin{array}{cccc}2& 2& 2& 1\\ 2& 1& -2& -2\\ 2& -1& -2& 2\\ 2& -2& 2& -1\end{array}\right].$

2. The method of claim 1, further comprising:

(a) dividing the output 4×

4 block by 2⁸.

3. The method of claim 1, wherein:

(a) said input 4×

4 block Y is of the form M X M^T for a 4×

4 block X of integers and M is the 4×

4 matrix;

$M=\left[\begin{array}{cccc}1& 1& 1& 1\\ 2& 1& -1& -2\\ 1& -1& -1& 1\\ 1& -2& 2& -1\end{array}\right].$

4. A method of N×

N blcok transformation, comprising;

(a) providing an input N×

N block of integer elements where N is an integer greater than 1;

(b) computing a first matrix product of said input N×

N block with a first N×

N matrix, said first N×

N matrix having integer elements;

(c) computing a second matrix product of said first matrix product with a second N×

N matrix, said second N×

N matrix having integer elements and being an integer approximation of the transpose of a third N×

N matrix wherein said third N×

N matrix is a fractional multiple of said first N×

N matrix; and

(d) normalizing said second matrix product.

5. The method of claim 4, wherein:

(a) said first N×

N matrix is an orthogonal matrix.

6. The method of claim 4, wherein:

(a) N equals 4; and

(b) said integer elements have 16-bit precision.

7. A method of N×

N integer block transformation, comprising;

(a) providing an input N×

N block of integer elements;

(b) computing a first matrix product of said input block with a first N×

N matrix, said first N×

N matrix with integer elements;

(c) normalizing said first matrix product;

(d) computing a second matrix product of said normalized first matrix product with a second N×

N matrix, said second N×

N matrix being a transpose of said first N×

N matrix; and

(e) normalizing said second matrix product.

8. The method of claim 7, wherein:

(a) said first N×

N matrix is an orthogonal matrix.

9. The method of claim 7, wherein:

(a) N equals 4; and

(b) said integer elements have 16-bit precision.