基于SiT-ROMP算法的视频封装帧压缩重构研究

摘要

Firstly,a design scheme of Encapsulating Video Frame ( EVF) is proposed,aiming at ameliorating the situation where the effi-ciency of reconstruction system is low because of incomplete correlation between video frame. Hence it takes full use of correlation of vid-eo frame between compression and reconstruction,what’ s more,its feasibility and the scientificity of improving the system performance are proved by theories and experiments. Then,according to disadvantages of compressing big data technology by current algorithms,the new improved algorithm named Self-iterative Threshold Regularized Orthogonal Matching Pursuit ( SiT-ROMP) is proposed. This algo-rithm not only guarantees the good quality of compression and reconstruction on video,but also shortens the reconstruction time greatly. And the experiment results show its feasibility and efficiency compared with other algorithms. Finally,combined EVF scheme and SiT-ROMP algorithm,propose a new method of SiT-ROMP algorithm based on EVF scheme. And the experiment results show that this scheme has considerable rationality and superiority in the quality and efficiency of compression and reconstruction on video.%文中首先针对由于视频帧间相关性利用不彻底而导致系统重构效率较低的现象，提出了视频封装帧设计方案，从压缩端和重构端双重考虑视频帧间相关性，并通过理论和仿真实验证明了封装帧方案的可行性和提高系统性能的科学性。其次针对目前重构算法对视频大数据压缩处理的先天不足，提出了自迭代阈值正则化正交匹配追踪( SiT-ROMP)算法，该算法既保证了很好的视频压缩重构质量，又大大缩短了重构时间，同时，通过仿真实验说明了SiT-ROMP算法的可行性以及相比其他重构算法的高效性。最后，将视频封装帧方案和SiT-ROMP算法相融合，提出了基于SiT-ROMP算法的视频封装帧压缩重构技术，并通过仿真实验表明了基于SiT-ROMP算法的视频封装帧压缩重构技术在视频压缩重构的质量和效率两个性能方面的合理性和优越性。