A Novel Video Compression Approach Based on Underdetermined Blind Source Separation

TL;DR

This paper introduces a new video compression method leveraging underdetermined blind source separation, significantly improving compression ratios when combined with existing codecs like MPEG-2 and H.264, while maintaining acceptable quality.

Contribution

It presents a novel integration of blind source separation with standard codecs to enhance video compression efficiency.

Findings

Over 33% improvement with MPEG-2

4X to 12X higher compression with H.264

Maintains acceptable PSNR levels

Abstract

This paper develops a new video compression approach based on underdetermined blind source separation. Underdetermined blind source separation, which can be used to efficiently enhance the video compression ratio, is combined with various off-the-shelf codecs in this paper. Combining with MPEG-2, video compression ratio could be improved slightly more than 33%. As for combing with H.264, 4X~12X more compression ratio could be achieved with acceptable PSNR, according to different kinds of video sequences.