ABSTRACT

There is shortage of novel methodologies in obtaining a better quality compressed video. Diverse exploration resulted in Group of pictures aiming at Intra, Bi-directional & Predicted frame identification for performing compression. The conventional GoP practices fixed frame pattern irrespective of the movement by the camera. This paper deals with the choice of adaptive frame identification by incorporating fixed and empirical values. The fixed values are chosen based on the various motion types. NSEW affine translation is incorporated to substitute B frames obtained from fixed and empirical threshold values with I or P frame. Motion Estimation & warping using affine are preferred for compression and decompression to obtain the compressed data and the warped. Four metrics explicitly have been analyzed to demonstrate the superiority of the suggested procedure.

摘要

在获得更好质量的压缩视频方面缺乏新的方法。多样化的探索导致了针对帧内、双向和预测帧识别的图片组，用于执行压缩。传统的 GoP 采用固定的帧模式，而与相机的移动无关。本文通过结合固定值和经验值来处理自适应框架识别的选择。固定值是根据各种运动类型选择的。NSEW 仿射平移被合并以用 I 或 P 帧替换从固定和经验阈值获得的 B 帧。使用仿射的运动估计和翘曲是压缩和解压缩的首选，以获得压缩数据和翘曲。