A Multi-View Multi-Description Video Coding Method Based on Data Reuse

Abstract

The invention discloses a multi-viewpoint multi-description video encoding-decoding method based on data reusing. According to the characteristics that four subsequences which are produced by spatial horizontal and perpendicular sampling according to multi-viewpoint video sequences are extremely close, every two of the four subsequences are combined to form two descriptions, one subsequence in each description is directly encoded by a standard multi-viewpoint video encoder during encoding, and prediction encoding is carried out on each macro block in the other subsequence in each description in three different data reusing modes according to the different properties of the macro block. The method provided by the invention is simple and easy to realize. Under the condition of improving the fault-tolerant capability of multi-viewpoint video encoding, high encoding efficiency is kept, and encoding calculation complexity is greatly reduced.

Description

technical field

[0001] The invention relates to the field of video coding and decoding in video communication, in particular to a multi-viewpoint and multi-description video coding method based on data reuse, which can be applied to video communication in unreliable channels. Background technique

[0002] With the advent of the mobile broadband 4G era, the Internet and wireless channels will become the main transmission channels for multi-viewpoint video. Data errors in wireless channel transmission, excessive delay in data arrival time and packet loss caused by channel interference in the Internet, network congestion and routing delay will seriously affect the reconstruction quality of multi-view video images. The existing multi-view video coding (Multi-view Video Coding, MVC) standard adopts the inter-view / temporal hybrid prediction structure based on hierarchical B frames proposed by HHI, which not only uses intra-frame and intra-view inter-frame prediction, but also Int...

Examples

Embodiment Construction

[0033] Such as figure 1 , figure 2 As shown, the present invention proposes a data reuse-based multi-view multi-description video encoding and decoding method, wherein,

[0034] On the encoding side, implement the following steps for each image group (GroupofPicture, GOP) of the input multi-view video:

[0035] Step 11. Carry out horizontal downsampling and vertical downsampling to each frame of each viewpoint of the input multi-viewpoint video image group to obtain F with the same resolution. 1p , F 1d , F 2p , F 2d Four sub-frames, and then F for each viewpoint 1p , F 1d , F 2p and F 2d Combined in frame order to form the four video subsequences V of this viewpoint 1p ,V 1d ,V 2p and V 2d ; Then the video subsequence V of each viewpoint 1p A collection of multi-view video subsequences X 1p The current image group, the video sub-sequence V of each viewpoint 1d A collection of multi-view video subsequences X 1d The current image group, the video sub-sequence V...

technical field

[0001] The invention relates to the field of video coding and decoding in video communication, in particular to a multi-viewpoint and multi-description video coding method based on data reuse, which can be applied to video communication in unreliable channels. Background technique

[0002] With the advent of the mobile broadband 4G era, the Internet and wireless channels will become the main transmission channels for multi-viewpoint video. Data errors in wireless channel transmission, excessive delay in data arrival time and packet loss caused by channel interference in the Internet, network congestion and routing delay will seriously affect the reconstruction quality of multi-view video images. The existing multi-view video coding (Multi-view Video Coding, MVC) standard adopts the inter-view / temporal hybrid prediction structure based on hierarchical B frames proposed by HHI, which not only uses intra-frame and intra-view inter-frame prediction, but also Int...

Examples

Embodiment Construction

[0033] Such as figure 1 , figure 2 As shown, the present invention proposes a data reuse-based multi-view multi-description video encoding and decoding method, wherein,

[0034] On the encoding side, implement the following steps for each image group (GroupofPicture, GOP) of the input multi-view video:

[0035] Step 11. Carry out horizontal downsampling and vertical downsampling to each frame of each viewpoint of the input multi-viewpoint video image group to obtain F with the same resolution. 1p , F 1d , F 2p , F 2d Four sub-frames, and then F for each viewpoint 1p , F 1d , F 2p and F 2d Combined in frame order to form the four video subsequences V of this viewpoint 1p ,V 1d ,V 2p and V 2d ; Then the video subsequence V of each viewpoint 1p A collection of multi-view video subsequences X 1p The current image group, the video sub-sequence V of each viewpoint 1d A collection of multi-view video subsequences X 1d The current image group, the video sub-sequence V...

Images