SVM Based Temporal Compression Techniques for Video Compression

Abstract

Compression is an important role in multimedia communication. In this paper, the problems of compression, quality degradation, best matched video, and image are discussed. Block matching algorithms compress the image block by block, and it also operates in a temporal manner. Proposed method is a SVM based algorithm compression. It is the combination of intensity and motion estimation. The proposed technique's compression gains of 92% accuracy with respective to static JPEG-LS, as it is applied in a frame basis. It achieves a good computation complexity and reduces it. An experimental result shows the better performance than existing methods. The method requires less best match points compared to existing methods. The quality of compression is improved when compared to the previous methods. Keywords Support vector machine (SVM) • Support vector decomposition (SVD) • Hue saturation intensity (HSI) • Region of interest (ROI) • Probability error (PE)