Font Size: a A A

Error Concealment Techniques For HEVC Video Based On Spatial-temporal Characteristics

Posted on:2018-01-03Degree:MasterType:Thesis
Country:ChinaCandidate:G B FanFull Text:PDF
GTID:2348330536479533Subject:Signal and Information Processing
Abstract/Summary:
High Efficiency Video Coding(HEVC)is the newest video coding standard for high and ultra-high definition video.HEVC outperforms the existing H.264/AVC high profile by 50% bitrate reduction at the same reconstructed video quality,causing HEVC coded videos are more vulnerable in transmission.Thus it is crucial to develop effective error control algorithm for HEVC.Error Concealment(EC)algorithm based on decoder is a focus of error control algorithm because it has many advantages.In this thesis,the EC is researched and the coding characteristics of HEVC are explored in different situations.The EC for HEVC is proposed from three angles.The main work and innovation are as follows:(1)An EC based on PU motion vector extrapolation(MVE)is proposed for the shortcomings of the existing error concealment algorithm for the loss of the whole frame of HEVC.First,establishing MVE algorithm appling to the HEVC.Then,PU does motion vector extrapolation for the lost frame,the LCU of lost frame is overlapped with the PU of the different split decision,and the split decision of the lost LCU is determined by the smallest PU split decision,thereby determining the ECBU of the lost LCU;then do the PU motion vector extrapolation for each ECBU.Repeat this process until the entire lost frame is concealed.(2)For the case of HEVC block loss,a weighted BMA algorithm based on PU merge is proposed.The algorithm first assumes that the CU/PU segmentation of the lost LCU is assumed to be the same as its co-located block;then,concerning that MV resulting from RDO may not represent the real motion of the video object,according to co-located block’s residual energy to determine whether the current split decision is reliable,and the adjacent unreliable PU within same CU depth merge as one,to refine the loss LCU’s segmentation;Next establishing the BMA based on the PU for the HEVC standard by the weight graph.After a PU is concealed,the weight graph must be updated,repeat this process until the entire lost LCU is concealed.(3)Aiming at the importance of video content in EC,an error concealment algorithm based on video content characteristics analysis is proposed for HEVC.By analyzing the LCU partitioning characteristic of HEVC,the algorithm can estimate the depth of the lost LCU by the coding depth information of correctly receiving spatial-temporal djacent LCU.If the frame area of lost LCU can be judged as simple and smooth,and the lost LCU can be concealed by a simple error concealment method.That is,the MV can be obtained by averaging the motion vector of the adjacent LCU.On the other hand,according to the characteristics of the HEVC for the intra mode,the ratio of the number of CU for the intra-frame and the total number of CU is calculated.If the frame area of the lost LCU belongs to the very intense motion area.At this time,the spatial concealment algorithm based on the intra prediction method is proposed by HEVC intra prediction method.The remaining frame area adopts the EC based on PU merge can achieve good error concealment effect.In the end,the summary of this study,as well as possible directions for future work,have been presented.
Keywords/Search Tags:HEVC, Error Concealment(EC), MVE, PU, BMA, analysis of content characteristics
Related items