Algorithm optimisation and hardware implementation of interprediction mode decision

High efficiency video coding is the most widely used video coding standard. It has higher coding performance compared with its predecessor, H.264, but it also has higher computational complexity. Interprediction is the most computationally intensive part of the entire video encoding process. Selecti...

Full description

Saved in:
Bibliographic Details
Published in:Journal of real-time image processing 2021-06, Vol.18 (3), p.593-601
Main Authors: Shi, Long-zhao, Yan, Danyu, Hong, Xiaojian, Huang, Bo, Yang, Xiuzhi
Format: Article
Language:English
Subjects:
Algorithms
Coding
Complexity
Computer Graphics
Computer Science
Distortion
Hardware
Image Processing and Computer Vision
Multimedia Information Systems
Original Research Paper
Pattern Recognition
Signal,Image and Speech Processing
Software
Video compression
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:High efficiency video coding is the most widely used video coding standard. It has higher coding performance compared with its predecessor, H.264, but it also has higher computational complexity. Interprediction is the most computationally intensive part of the entire video encoding process. Selecting the optimal interprediction mode by the rate–distortion cost calculation function requires substantial complex calculation and memory access, thus greatly increasing the difficulty of real-time hardware encoding. This study proposes to replace the traditional complex error square sum calculation with an estimation method for distortion and rate. The estimation of distortion uses the Hadamard-transformed sum of absolute transformation difference instead of the complex calculation of the sum of squared difference, whereas the estimation of rate is obtained by weighting the number of prediction units (PUs). The experiment proves that the proposed interprediction rate–distortion cost calculation model can greatly reduce computational complexity when BD-rate is increased by 3.02%. In hardware implementation, the value of rate can be obtained by indexing the number of PUs, and the resource expenditure is small.
ISSN:1861-8200
1861-8219
DOI:10.1007/s11554-020-00985-w