Research On Fast Mode Selection Algorithm For Intra Prediction Of HEVC

With the rapid development of Internet,multimedia,electronic chip technology,improved shooting digital video camera can provide users with a high resolution has been from the traditional 720P to 2K,4K.In Ultra HD video era,the video capacity is growing while results in a lack of bandwidth resources.In order to meet the real-time data transmission of video surveillance need for macional security and finamial bank,scientists put forward the HEVC technology which can do the video compression encoding and reduce video capacity.In HEVC video compression encoding process,in order to improve the encoding efficiency,we can use the key technologies of intra prediction which can make full use of temporal correlation between video flow to improve the HEVC encoding process.It can as much as possible reduce the rate distortion cost and maintain the quality of the video image,and in the same time it can reduce the video compression encoding time.This paper proposes a fast HEVC intra prediction mode selection algorithm,the main work includes three aspects listed as follows:Firstly,the paper provided detailed description of the HEVC video encoding process and analyzed the video encoding framework of HEVC,and the classification of HEVC image encoding.The encoding and encoding unit types included tree tree structure design and the encoding unit and encoding block,forecast and prediction unit block and block transform unit.There are a lot of key technology of HEVC in the process of encoding used.Indigital video which are HEVC intra prediction,inter prediction,HEVC transform and quantization,entropy encoding,deblocking filter and adaptive sample migration techniques.Those techniques can provide technical support for the prediction of HEVC encoding.The paper analyzed the related technologies to improve the effectiveness of HEVC coding,which techniques can optimize the HEVC effect,and those technologies include the SAD cost function,the STAD cost function,the rate distortion cost function and the RQT technology.Secondly,the paper summarized the existing problems of HEVC frame prediction technology,and selectively analyzed the problems of low efficiency and poor quality compression in Hammond transform and coarse mode selection of video compression.We propose an adaptive frame grouping fast intra prediction mode selection algorithm based on the basic idea of data mining clustering.The new algorithm can better mine the special correlation of video image compression encoding process,and it can better delete candidate patterns after rough selection which can better reduce the aging video image compression encoding.Adaptive packet algorithm running mainly includes two key stages:(1)Choosing the candidate pattern after rough data mining clustering group,according to the different model groups,the relevant edge prediction processing method can be seleted to fit candidate RDO mode;(2)Carrying out analyses for many kinds of prediction units,such as the 16x16,32×32,64×64 and so on,the RDO candidate mode number can be greatly reduced,and it greatly reduced the time of HEVC intra prediction.Finally,in order to run the results validate the algorithm,the paper adopted HEVC official test version of HM16.7 to analyze the results of the experiment.The experimental results showed that the proposed HEVC algorithm based on adaptive grouping which maintains the image quality can greatly reduce the time of the intra frame prediction encoding,and it is suitable for all kinds of images with different size and properties the image.