Video Compression Algorithm And Key Issues Based On H.264/AVC

Along with the rapid development of4G communications and multimediaservices, the high-efficiency video compression coding technology has been obtainedmore and more attention. Although network bandwidth and storage capacity has beenmade great strides in recent years, it is still far from being able to meet the soaringdemand for the video transmission and storage space characterized by the massivedata, so the high-performance video coding compression technology is bound to be animportant and indispensable research hotspot. Due to its superior compressionperformance compared to older codecs, H.264/AVC has penetrated into every aspectof our lives and become the most ubiquitous video compression standard. As one ofthe most highest-tech video coding formats, H.264/AVC represents the latest technicallevel, a combination of market expansion, technology advances and increased userexpectation make the compression algorithm research in this field of great practicalsignificance.This paper makes a study of the relevant theories about the H.264/AVC videocompression standard, and theoretical research is divided into two parts: one is thealgorithm optimization of the inter prediction skip mode; another is the motionestimation search template improvement based on the UMHexagonS algorithm. Afterintroducing the main technology and feathers, chapter three put forward a new skipmode judgment strategy to early terminate the following process by adjusting andrelaxing the restrictions which could further remove the spatial redundancy based onthe strong correlation of the adjacent blocks and the current block as well as therelatively low concern about boundary portion of video images; chapter four presentsa novel search window template combined with the recommend UMHexagonSalgorithm. Through utilizing the directional characteristics of the motion vector, thesearch pixel points of the non-uniform multi-hexagon search and the extendedhexagon search window can be simplified greatly, and the motion estimation timehave be shorten.