Research On Intra Prediction Algorithm Of HEVC For Frame Of Wavelet Transform

As the latest video compression standard,H.265/HEVC has attracted much attention because of its high compression efficiency and excellent compression effect.Compared with previous H.26 X standards,H.265/HEVC selects more intra prediction modes and finer prediction units,which makes HEVC have higher coding efficiency and better compression effect..However,with the improvement of coding efficiency,the computational complexity of encoding increases exponentially,which makes real-time encoding and decoding of HEVC very difficult and hinders the popularization and use of HEVC.The rate distortion optimization rule encoding mode selection more time spent encoding based on 82%,therefore,how to reduce the encoding complexity of HEVC,especially on how to reduce the number of encoding unit,division of intra mode selection,thereby reducing the computation time has become the key to the optimization of HEVC compression encoding algorithm.Since the wavelet transform has the characteristics of high frequency sub-band directionality and more concentrated distribution of texture information,it makes it possible to use these features to guide HEVC intra prediction.In this paper,a series of experiments are carried out to verify the correlation between the subband texture and the intra prediction mode,and the correlation between the texture complexity and the encoding unit partition.Based on the above analysis,a reasonable texture detection scheme is designed.Finally,the intra prediction mode fast selection algorithm for DWT frames is designed based on the correlation between texture direction and intra prediction mode and texture detection scheme.The algorithm increases the coding speed by more than 35% while reducing the bit rate(with high accuracy).A fast encoding algorithm for DWT frames is designed based on the correlation between the texture complexity and the coding unit.The method increases the coding speed by morethan 40% while reducing the bit rate(with high accuracy).Integrating the above two algorithms and reusing the texture detector at the same time,the coding speed is raised by more than 50% on the basis of reducing the 3.44% bit rate(with higher accuracy).