Browse Prior Art Database

Directional Discrete Cosine Transform (DDCT) for intra coding in video compression applications

IP.com Disclosure Number: IPCOM000199424D
Publication Date: 2010-Sep-03

Publishing Venue

The IP.com Prior Art Database

Related People

Thai-Ha Nguyen: INVENTOR [+3]

Abstract

We propose a technique to code intra-frames which can be used in video compression applications. The technique consists of a set of 4x4 and 8x8 transforms that will be applied on the intra prediction errors.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 26% of the total text.

Sony Corporation

Sony Electronics Inc.

IPD Case # 200903014

Title:

Directional Discrete Cosine Transform (DDCT) for intra coding in video compression applications

Description of the Invention:

We propose a technique to code intra-frames which can be used in video compression applications. The technique consists of a set of 4x4 and 8x8 transforms that will be applied on the intra prediction errors.

Inventor:

Thai-Ha Nguyen

Wei Liu

Ali Tabatabai

Invention Disclosure:

What issues/problems does your invention try to address?

We try to find better transforms to be applied on intra prediction errors in the coding of intra frames in video compression technology H.264. The goal is to make the transform coefficients sparse so that they can be easily coded and sent to the decoder.

How did others try to address these issues/problems?

The current transform uses DCT horizontally and vertically, hence across edges. This make the transform not optimal as well as the visual quality of the reconstructed image is affected.

Explain your invention and specifically identify the new and novel features.

We apply the transform along and across the prediction direction instead of horizontal and vertical. DCT of different sizes are used.

1.     The DDCT transform

The Directional Discrete Cosine Transforms (DDCT) is a set of transform to apply to the intra prediction errors in the video compression framework AVC/ H.264. In this section, description and properties of DDCT will be described.

1.1.  Intra coding in AVC/ H.264 and where DDCT fits in:

Intra coding predicts the image content based on the value of previously decoded pixels. It has 9 prediction modes for 4x4 blocks, 9 prediction modes for 8x8 blocks, and 4 prediction modes for 16x16 blocks. For each intra prediction mode, an intra prediction algorithm is used to predict the image content in the current block based on decoded neighbors. The intra prediction errors are transformed using a DCT-like transform. The transform coefficients are then quantized, scanned into a 1D signal, and entropy coded using CAVLC or CABAC.

In this framework, DDCT is to replace the AVC/ H.264 transforms by a set of transforms that taking into account the prediction mode of the current block. Hence, DDCT provides 9 transforms for 4x4, 9 transforms for 8x8, and 4 transforms for 16x16, although many of them are the same or can be simply inferred from a core transform. For each transform, the DDCT also provides a fixed scanning pattern based on the QP and the intra prediction mode to replace the zigzag scanning pattern of DCT coefficients in AVC/ H.264.

1.2. Directional Discrete Cosine Transform (DDCT):

Transforms: DDCT provide 9 transforms for 4x4, 9 transforms for 8x8, and 4 transforms for 16x16. For each intra prediction mode, DDCT transform consists of two stages:

-          Stage 1 – along the prediction direction: pixels that align along the prediction direction are grouped together and feed into a DCT transform. Note that, in cases of predicti...