Research On Error Concealment Techniques Of Image Coding Transmission

With the increasingly high demand of image quality and resolution, video compressionhas become an essential way to store and transport information. As next-generationcommunication standard of video coding, H.264/AVC not only has a high compression ratiobut also the quality of restoring images is higher than using the traditional encoding methods.Coding uses motion compensation, prediction and variable length coding, etc. to removethe video images’ correlation in the temporal and spatial. Video images obtaining a loweramount of redundancy and higher compression will make its error resilience weak afterencoding. If we don’t take any technical control when an error occurs, damaged imagedecoding in the decoder can cause errors spread to the entire video sequence frames.According to the characteristics of the video compression algorithms and transmissionchannel, studying error control techniques becomes an important research of video codingtransmission technology in recent years.For the problem that restoration quality of images existing rich edge information is low bytraditional spatial error concealment algorithm of H.264/AVC, this paper presents a spaceerror concealment algorithm based on adaptive edge threshold and direction weighting. Firstly,this algorithm detects neighboring macro-block edgily, and then according to the specificinformation of its neighboring macro-block sets gradient threshold adaptively, at last thedamaged macro-block is reconstructed by direction weighted interpolation. Experiments showthat the algorithm is simple and practical. In different probability events of RTP packet losing,the PSNR has improved. This algorithm not only improves the quality of image restoration,but has higher application value.The important problem of time-domain error concealment is reconstructing the motionvectors of damaged macro-blocks. Based on the motion vector of neighboring macro-blocksand the zero motion vector, traditional error concealment algorithm uses boundary matchingalgorithm to select the best motion vector and achieves error concealment ultimately. For theproblem of motion vector mismatch of the traditional time-domain error concealment algorithm in H.264/AVC, this paper presents an improved error concealment algorithm. Thisalgorithm reconstructs the motion vector of the damaged macro-block by the fitting planealgorithm, and uses the outer boundary matching algorithm to select the best motion vector toreconstruct the damaged images. This algorithm not only avoids the block effects that causedby the original algorithm, but also in different probability of RTP packet loss, the peak signalto noise ratio of the algorithm has been increased and there is a big improvement on thequality of images that restored from the subjective aspect and objective aspect.