Research On Screen Content Coding Based On H.265/HEVC

In recent years,screen content has been widely used in distance education,multimedia presentation,cloud computing and other fields.The huge difference between screen content and camera-captured content makes it difficult for the traditional encoder to compress screen content.In order to improve the coding efficiency for screen content,the high efficiency video coding standard H.265/HEVC has introduced the screen content coding tools,according to the features of screen content such as irregular motion,large motion,and large uniformly flat areas.The encoding algorithms contain a series of new coding techniques.In this paper,the screen content coding methods are studied,and a series of algorithms are proposed to improve the compression performance on screen content.And the research is divided into the following two aspects:First,we propose an inter prediction scheme based on rubost hash.In inter prediction procedure,the encoder matches the reference block which predicts the current block with the smallest distortion.However,the hash-based inter prediction method of the encoder in the HEVC-SCC frame can not match blocks which are similar with the current block.Thus,we propose an inter prediction scheme based on rubost hash,which uses two hash functions to match blocks.The first hash function matches the blocks which are identical to the current one;and the second hash function is used to match the blocks which are similar with the current one.In order to reduce the complexity,we also design fast algorithms to avoid repeated calculation while calculating the hash values.Finally,the proposed scheme is embedded in the inter-predictioin procedure of HEVC-SCC.Experimental results show that the inter prediction scheme based on rubost hash improves the coding performance,compared with the original scheme in HEVC-SCC.Secondly,we propose a screen content codign algorithm based on optimal picture reordering.The irregular motion of objects weakens the time correlation of screen content.In order to enhance the time correlation of screen content,we propose a screen content codign scheme based on optimal frame reordering.The screen content video frames are divided into groups and the images within the group are reordered with the order of groups unchanged.In the scheme,the hash value of the image block is obtained by using the hash function proposed in the inter prediction scheme based on rubost hash.Then,using the hash-based method,the correlation between images are calculated.Next,according to the correlation between images and the parameter H.265/HEVC uses in the screen content coding,the images are reordered.In order to further improve the coding efficiency,a reference picture selecting algorithm is implemented.Finally,the scheme is embedded in the HEVC-SCC framework for experiments.Experimental results show that the coding performance of the proposed scheme is greatly improved compared with the original scheme in the HEVC-SCC.