
Research Article
An Efficient Compression Coding Method for Multimedia Video Data Based on CNN
@INPROCEEDINGS{10.1007/978-3-031-28787-9_10, author={Xu Liu and Yanfeng Wu}, title={An Efficient Compression Coding Method for Multimedia Video Data Based on CNN}, proceedings={Advanced Hybrid Information Processing. 6th EAI International Conference, ADHIP 2022, Changsha, China, September 29-30, 2022, Proceedings, Part I}, proceedings_a={ADHIP}, year={2023}, month={3}, keywords={CNN Multimedia video Video data Compression coding Video coding Efficient compression}, doi={10.1007/978-3-031-28787-9_10} }
- Xu Liu
Yanfeng Wu
Year: 2023
An Efficient Compression Coding Method for Multimedia Video Data Based on CNN
ADHIP
Springer
DOI: 10.1007/978-3-031-28787-9_10
Abstract
Due to the phenomenon of object occlusion and inconsistent motion of different objects in video, high-efficiency compression and coding of video data will generate prediction residuals related to texture structure. To solve this problem, this study proposes an efficient compression coding method for multimedia video data based on CNN. First, the multimedia video coding unit is divided, and the coded frames are arranged in POC order for coding. Then, the coding structure adjustment parameters are calculated, and after coding, the determined reference frame and the bit consumption caused by using the reference frame can be obtained. Finally, an intra-prediction algorithm for video data is established based on CNN. CNN encoder uses a series of down-sampling convolution and ReLU nonlinear mapping to extract and fuse global information, and conducts analysis on areas with low human visual sensitivity on the HEVC transform domain. Frequency coefficient suppression. The experimental results show that the method has good coding performance for test sequences with different contents and different resolutions.