32-point transform for media data coding

Patent metadata

Field	Value
Publication number	US-9824066-B2
Application number	US-201213346632-A
Country	US
Kind code	B2
Filing date	Jan 9, 2012
Priority date	Jan 10, 2011
Publication date	Nov 21, 2017
Grant date	Nov 21, 2017

Abstract

In general, techniques are described for implementing a 32-point discrete cosine transform (DCT) that is capable of applying multiple DCTs of different sizes. For example, an apparatus comprising a 32-point discrete cosine transform of type II (DCT-II) unit may implement the techniques of this disclosure. The 32-point DCT-II unit performs these DCTs-II of different sizes to transform data from a spatial to a frequency domain. The 32-point DCT-II unit includes an 16-point DCT-II unit that performs one of the DCTs-II of size 16 and at least one 8-point DCT-II unit that performs one of the DCTs-II of size 8. The 16-point DCT-II unit includes another 8-point DCT-II unit. The 16-point DCT-II unit also comprises at least one 4-point DCTs-II unit. Two or more of these DCTs-II units may concurrently perform DCTs-II of different sizes to various portions of the content data.

First claim

What is claimed is:

1. A method for transforming video data between a spatial domain and a frequency domain, the method comprising:
performing one or more of multiple scaled discrete cosine transforms of type II (DCTs-II) of different sizes with a scaled 32-point DCT-II transform unit of a video coding device to transform a portion of the video data between the spatial domain and the frequency domain, wherein the scaled 32 DCT-II transform unit performs a first one of the plurality of scaled DCTs-II of size 32, wherein the multiple scaled DCTs-II include 32 internal fixed-point factors selected such that each of the multiple scaled DCTs-II satisfies an orthogonality property, wherein the scaled 32-point DCT-II transform unit includes:
a scaled 16-point DCT-II transform unit located in an even half of the scaled 32-point DCT-II transform unit that performs a second one of the multiple scaled DCTs-II, the second one of the multiple scaled DCTs-II defining a transform size of 16;
a first scaled 8-point DCT-II transform unit located in an odd half of the scaled 32-point DCT-II transform unit that performs a third one of the multiple scaled DCTs-II, the third one of the multiple scaled DCTs-II defining a transform size of 8;
a second scaled 8-point DCT-II transform unit located in the odd half of the scaled 32-point DCT-II transform unit that performs a fourth one of the multiple scaled DCTs-II, the fourth one of the multiple scaled DCTs-II defining a transform size of 8; and
a 16-point butterfly unit defined by a Givens rotation matrix located in the odd half of the scaled 32-point DCT-II transform unit.

2. The method of claim 1 , wherein performing the one or more of the multiple scaled DCTs-II comprises performing the second one of the multiple scaled DCTs-II of size 16 concurrent to performing the either the third one or the fourth one of the multiple scaled DCTs-II of size 8.

3. The method of claim 1 ,
wherein the scaled 16-point DCT-II transform unit includes:
a third 8-point DCT-II transform unit;
a first 4-point DCT-II transform unit; and
a second 4-point DCT-II transform unit,

wherein performing the one or more of the multiple scaled DCTs-II of different sizes comprises concurrently performing two or more of the multiple scaled DCTs-II of sizes 8 and 4 with one or more of the first, second, and third 8-point DCT-II transform units and one or more of the first and second 4-point DCT-II transform units.

4. The method of claim 3 ,
wherein performing the one or more of the multiple scaled DCTs-II of different sizes comprises:
performing one of the multiple scaled DCTs-II of size 8 either 1) without performing matrix multiplication or 2) using a combination of butterflies and matrix multiplication; and
performing one of the multiple scaled DCTs-II of size 4 either 1) without performing matrix multiplication or 2) using a combination of butterflies and matrix multiplication.

5. The method of claim 3 , wherein performing the one or more of the multiple scaled transforms of different sizes comprises:
concurrently performing one of the multiple scaled DCTs-II of size 4 and one of the multiple scaled DCTs-II of size 8; and
performing one of the multiple scaled DCTs-II of size 8 concurrent to performing the one of multiple scaled DCTs-II of size 4 and the one of the multiple scaled DCTs-II of size 8.

6. The method of claim 5 , wherein performing the one of the multiple scaled DCTs-II of size 4 comprises performing the one of the multiple scaled DCTs-II of size 4 either 1) without performing matrix multiplication or 2) using a combination of butterflies and matrix multiplication.

7. The method of claim 1 ,
wherein performing the one or more of the multiple scaled DCTs-II of different sizes comprises performing the one or more of the multiple scaled DCTs-II of different sizes with the scaled 32-point DCT-II transform unit to transform the first portion of the video data from the spatial domain to the frequency domain,
wherein the method further comprises:
outputting scaled transform coefficients representative of the first portion of the video data in the frequency domain; and
applying scale factors with a quantization unit to the scaled transform coefficients so as to generate quantized full transform coefficients.

8. The method of claim 7 , wherein applying the scale factors comprises applying the scale factors using a look-up table (LUT) included within the quantization unit that stores a mapping between the scaled transform coefficients and entries of a quantization matrix.

9. The method of claim 7 , wherein applying the scale factors comprises applying the scale factors based on a mapping between the scaled transform coefficients and entries of a quantization matrix with the quantization matrix of a size that is smaller than a full quantization matrix specifying scale factors for each of the scaled transform coefficients, and is dependent upon one or more of the 32 internal fixed-point factors.

10. The method of claim 1 , wherein performing the one or more of the multiple scaled DCTs-II comprises:
performing one of the multiple scaled DCTs-II of size 16 either 1) without performing matrix multiplication or 2) using a combination of butterflies and matrix multiplication, and
performing one of the multiple scaled DCTs-II of size 8 either 1) without performing matrix multiplication or 2) using a combination of butterflies and matrix multiplication.

11. The method of claim 1 , further comprising:
performing motion estimation with respect to a block of the video data to identify a reference block of the video data and generate motion vectors based on the identified reference block of video data;
performing motion compensation to determine a prediction block based on the motion vectors;
comparing the prediction block to the block of video data to generate a residual block of video data,
wherein performing the one or more of the multiple scaled DCTs-II of different sizes comprises performing the one or more of the multiple scaled DCTs-II of different sizes to transform the residual block of video data from the spatial domain to the frequency domain;
quantizing the transformed residual block of video data to generate a quantized block of video data; and
performing statistical coding to code the quantized block of video data and generate a coded block of video data.

12. The method of claim 1 , further comprising:
statistically decoding a coded block of the video data to generate a quantized block of video data;
performing inverse quantization on the quantized block of video data to generate a transformed block of video data,
wherein performing the one or more of the multiple scaled DCTs-II of different sizes comprises performing the one or more of the multiple scaled DCTs-II to the transformed block of video data so as to generate a residual block of video data; and
performing motion compensation on the residual block of video data to determine a decoded block of the video data.

13. The method of claim 1 ,
wherein the video coding device comprises a wireless communication device,
wherein the method is executable on the wireless communication device,
wherein the wireless communication device comprises a memory configured to store the video data,
wherein the scaled 32-point DCT-II transform unit comprises a processor configured to execute instructions to perform the one or more of the multiple scale DCTs-II of the different sizes to transform the portion of the video data between the spatial domain and the frequency domain.

14. The method of claim 13 ,
wherein the wireless communication device comprise

Assignees

• Joshi Rajan Laxman
• Reznik Yuriy
• Sole Rojals Joel
• Karczewicz Marta
• Qualcomm Inc

Inventors

• Joshi Rajan Laxman
• Reznik Yuriy
• Sole Rojals Joel
• Karczewicz Marta

Classifications

• H04N19/625

  using discrete cosine transform [DCT] · CPC title

• G06F17/14 Primary

  Fourier, Walsh or analogous domain transformations {, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms (for correlation function computation G06F17/156; spectrum analysers G01R23/16)} · CPC title

• G06F17/147 Primary

  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 title

• H04N19/60

  using transform coding · CPC title