Intelligent inter-viewpoint coding method

Abstract

The invention discloses an intelligent inter-viewpoint coding method. The method comprises the following steps: respectively obtaining a first prediction block obtained based on an adjacent block parallax vector mode and a second prediction block obtained based on a traditional parallax compensation prediction mode; constructing a residual learning convolutional neural network structure with a variable convolution kernel size; taking a result obtained by cascading the brightness component channels of the two prediction blocks as the input of a convolutional neural network, wherein the output of the network is an enhanced prediction block of the original image block; training four different network models according to the four different quantization parameters; extracting brightness components of the first prediction block and the second prediction block respectively, performing channel cascading, and calling a corresponding network model for a result after cascading according to the size of the current coding block and the quantization parameter value; and finally, calculating the rate distortion cost of the enhanced prediction block, comparing the rate distortion cost with the optimal mode cost of the current coding block; if the rate distortion cost is less than the optimal mode cost, selecting the method as the optimal coding mode, coding the corresponding mark, and transmitting the coded mark to a decoding end.

Description

technical field [0001] The invention relates to the field of 3D video coding, in particular to a coding method between intelligent viewpoints. Background technique [0002] The large amount of 3D video data brings unprecedented difficulties and challenges to video storage and transmission. Therefore, how to realize efficient 3D video coding has important theoretical research significance and practical application value. Currently, 3D-HEVC (3D High Efficiency Video Coding) developed by JCT-VC, an international video coding standard organization jointly established by MPEG (Motion Picture Experts Group) and VCEG (Video Coding Experts Group), is the latest 3D video coding standard. 3D-HEVC introduces new prediction technology and coding tools for non-independent viewpoint and depth video, adds technologies suitable for multi-view video coding and depth video coding, and improves coding efficiency. However, there is still a lot of room for improvement in inter-view prediction ...

AI Technical Summary

Problems solved by technology

[0002] The large amount of 3D video data brings unprecedented difficulties and challenges to video storage and transmission

[0007] The existing algorithms for modeling parallax relations, such as stretching and cutting the reference image, can effectively improve the accuracy of prediction, but the applicability is poor, and the effect is not good for videos with complex textures and rich movements; currently there is no Coding Method Between Intelligent Viewpoints

Images