Research On Video Lossless Compression Algorithm Based On HEVC

Currently, digital video and digital image processing technology has been widelyused in the field of communication, television broadcasting, storage and multimediaentertainment. However, due to the huge amount of video and image data, the storageand communication have been great challenged, especially in the field of security,medical, aerospace and so on, in which the video or image needs lossless or nearlylossless compression. But the lossless compression ratio is not satisfactory, therefore,it is an urgent task how to compress lossless video or image efficiently.This paper is based on the framework of High Efficiency Video Coding, referredas HEVC, which acts as the new generation video coding standard and has superiorperformance than the H.264/AVC. After we have researched the existing losslesscompression algorithms, we will improve the lossless compression performance ofvideo or image based on HEVC framework from two aspects: intra prediction andinter prediction.Firstly, we can use the HEVC framework itself to achieve splitting adaptivelycoding unit, determining the best segmentation scheme and selecting the optimalprediction mode based on the smoothness degree of image. The preliminaryprediction uses Differential Pulse Coding Modulation, referred as DPCM, which mayresult in large prediction error for the areas with complicated texture. So, the errorcompensation is used to correct the prediction values. Since the distributioncharacteristics of residuals depends on the prediction mode, which can determine thescanning order of residuals to make the entropy coding more efficient. Besides, weproposed hierarchical prediction, template matching, adaptive prediction and so on, asfar as possible to improve prediction accuracy, thereby reducing the spatialredundancy in order to improve the coding efficiency.Secondly, due to a large number of temporal redundancy in the video, the codingefficiency can considerably be improved by inter prediction. However, in the losslesscompression scheme, the transformation and quantization modules have beenremoved. So we increased spatial prediction on the basis of inter prediction and chosewhich mode to predict through the combination of temporal and spatial prediction.And after prediction, the prediction residuals were predicted once more, to furthereliminate redundancy and achieve the purpose of reducing the bit rate.