Method and apparatus for intra prediction in video coding system

Patent metadata

Field	Value
Publication number	US-10778985-B2
Application number	US-201615762958-A
Country	US
Kind code	B2
Filing date	Sep 23, 2016
Priority date	Sep 23, 2015
Publication date	Sep 15, 2020
Grant date	Sep 15, 2020

Abstract

An image decoding method, which is performed by a decoding apparatus, includes the steps of: deriving a frequency domain-neighboring block for a neighboring block of a current block in a spatial domain; deriving a frequency domain-prediction block for the current block by applying a prediction pattern to the frequency domain-neighboring block; generating a frequency domain-reconstructed block for the current block on the basis of the frequency domain-prediction block; and generating a reconstructed block for the current block in the spatial domain on the basis of the frequency inverse transform of the frequency domain-reconstructed block. The present invention can generate a prediction block for a current block in a frequency domain, and accordingly can more efficiently reconstruct a complex image.

First claim

What is claimed is:

1. An image decoding method performed in a video decoding apparatus, comprising steps of:
deriving a frequency domain neighbor block for a neighbor block of a current block in a spatial domain;
deriving a frequency domain prediction block for the current block by applying a prediction pattern to the frequency domain neighbor block;
generating a frequency domain reconstruction block for the current block based on the frequency domain prediction block; and
generating a reconstruction block for the current block in the spatial domain based on frequency inverse transform for the frequency domain reconstruction block,
wherein the frequency domain prediction block is derived by multiplying a frequency component of the frequency domain neighbor block by a value of a component within the prediction pattern at a location corresponding to the frequency component, and the value of the component within the prediction pattern is one of 0 and 1.

2. The image decoding method of claim 1 , wherein the neighbor block of the spatial domain comprises at least one of a top neighbor block, left neighbor block, top-left neighbor block, and top-right neighbor block of the current block.

3. The image decoding method of claim 1 , further comprising a step of obtaining index information indicative of the neighbor block of the spatial domain through a bit stream,
wherein the neighbor block of the spatial domain is determined to be one of neighbor blocks comprising a top neighbor block, left neighbor block, top-left neighbor block, and top-right neighbor block of the current block, and
the frequency domain neighbor block is derived based on the neighbor block belonging to the neighbor blocks and indicated by the index information.

4. The image decoding method of claim 1 , wherein the frequency domain neighbor block is derived as a block having a size identical with a size of the current block.

5. The image decoding method of claim 1 , wherein, based on a size of the frequency domain neighbor block and the prediction pattern being N×N and coordinates of a top-left component of the prediction pattern being (0,0), the prediction pattern is derived based on an equation below,

H
⁡

(

m
,
n

)

=

{

1
,

if
⁢

⁢
n

≤

N
/
2

0
,

otherwise

wherein m indicates coordinates of the prediction pattern in a vertical direction and n indicates coordinates of the prediction pattern in a horizontal direction.

6. The image decoding method of claim 1 , further comprising steps of:
configuring prediction pattern candidates based on a prediction pattern of the neighbor block; and
obtaining a prediction pattern index through a bit stream,
wherein the prediction pattern is derived based on a prediction pattern candidate belonging to the prediction pattern candidates and indicated by the prediction pattern index.

7. The image decoding method of claim 1 , further comprising steps of:
generating a most probable pattern (MPP) list based on a prediction pattern of a first neighbor block and prediction pattern of a second neighbor block of the current block in the spatial domain; and
obtaining an MPP index through a bit stream,
wherein the MPP list comprises prediction patterns, and
a specific prediction pattern belonging to the prediction patterns of the MPP list and indicated by the MPP index is derived as the prediction pattern of the current block.

8. The image decoding method of claim 7 , further comprising a step of obtaining a previous intra luma pattern flag from the bit stream,
wherein if a value of the previous intra luma pattern flag is 1, the MPP index is obtained through the bit stream.

9. The image decoding method of claim 8 , wherein:
based on the value of the previous intra luma pattern flag being 0, remaining intra luma prediction pattern information is obtained from the bit stream, and
the remaining intra luma prediction pattern information indicates one of remaining prediction patterns other than the prediction patterns of the MPP list.

10. The image decoding method of claim 1 , further comprising a step of obtaining a frequency domain intra flag through a bit stream,
wherein if a value of the frequency domain intra flag is 1, the frequency domain prediction block for the current block is derived.

11. The image decoding method of claim 1 , further comprising a step of obtaining frequency domain residual information for the current block through a bit stream,
wherein the frequency domain reconstruction block is generated based on the frequency domain prediction block and the frequency domain residual information.

12. An image encoding method performed in an image encoding apparatus, comprising steps of:
generating a frequency domain original block by performing frequency transform on a current block of a spatial domain;
deriving a frequency domain neighbor block for a neighbor block of the current block;
determining a prediction pattern of the current block;
deriving a frequency domain prediction block for the current block by applying the prediction pattern to the frequency domain neighbor block;
generating frequency domain residual information based on the frequency domain original block and the frequency domain prediction block; and
encoding information about a prediction mode of the current block and the frequency domain residual information and outputting the encoded information,
wherein the frequency domain prediction block is derived by multiplying a frequency component of the frequency domain neighbor block by a value of a component within the prediction pattern at a location corresponding to the frequency component, and the value of the component within the prediction pattern is one of 0 and 1.

13. The image encoding method of claim 12 , wherein:
the frequency domain residual information is a difference value between samples according to phases of the frequency domain original block and the frequency domain prediction block, and
the prediction pattern is a prediction pattern deriving a frequency domain prediction block in which an amount of data of the frequency domain residual information is minimized.

14. The image encoding method of claim 12 , wherein, based on a size of the frequency domain neighbor block and the prediction pattern being N×N and coordina

Assignees

• Lg Electronics Inc

Inventors

• Heo Jin
• Son Eunyong
• Nam Junghak
• Park Seungwook
• Yea Sehoon

Classifications

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/593

  involving spatial prediction techniques · CPC title

• H04N19/159 Primary

  Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction · CPC title

• H04N19/11

  among a plurality of spatial predictive coding modes · CPC title

• H04N19/61

  in combination with predictive coding · CPC title