Study On Video Data Hiding And Its Application In Error Concealment

With the rapid development of multimedia communication and intelligent terminal technologies,video has become the main media of multimedia communication due to its good visibility.Compared with images,video sequences have much abundant temporal and spatial correlation.Transmitting original video sequence requires great bandwidth,therefore,it needs video encoding /decoding standard,which takes advantage of the correlation of intra-frame and inter-frame,to remove redundancy information to encode or decode videos.The encoded bit stream is sensitive to errors,and small errors may result in the video quality degradation.In channel,weather packet lost and bit error occurs is decided by transmission mechanism and channel state.So,the quality of transmitted video depends on the robustness of the bit-stream and error control mechanism.The data hiding-based error concealment schemes are used to embed information which reflects frame feature at the encoder side.In the error concealment scheme,the more information which reflect the frame feature is embedded,the better concealed video quality.So,it is necessary to research the data hiding scheme with better capacity-distortion performance,as well as the video error concealment.This paper focuses on the research of video data hiding and its application in error concealment,and to provide the solution for optimization problem of minimizing embedding distortion and improving error concealment quality.The innovations are as follows.1.An adaptive video data hiding through cost assignment and STCs is proposed.Considering the statistical characteristics and the complexity of encoding/decoding standards,there are few studies about video data hiding schemes,and even fewer of them is based on embedding cost assignment.Based on the intra-frame spatial frequency characteristics and the temporal correlation,embedding cost assignment function is constructed.While,inter-frame correlation consists of motion feature and reference relationship between neighbor frames,and we use motion vector and frame order to indicate them.When data is embedded using STCs in the cover in which elements are selected with the minimized impact on subsequent frames in inter-frame encoding,embedding costs are assigned.Meanwhile,considering the frame encoding order,and the previous encoded frames are reference of the posterior frames,and the data is embedded into the posterior frames adaptively to perform the cost assignment–based video data hiding to reduce error iteration.2.A video error concealment method using data hiding based on compressed sensing is proposed.For video,motion vectors are two-dimensional data with low redundancy and re-flect the motion trace,and motion residuals are a main entity of encoding and transmitting in the video communication.That means that motion residual and motion vector are the reflection of inter-frame correlation.Motion residuals are difference between current frame/block and the reference frame/block,and reference blocks are selected by the best matching principle,so,the motion residuals are usually sparse and redundant.Considering the low sampling rate and better reconstruction performance of compressive sensing(CS),the motion residuals are compressed by CS as marked data.At the same time,in order to resist the random transmission error,on the one hand,the motion residuals of all frames in a GOP(Group of Picture)are compressed and considered as marked data.On the other hand,to prevent inter-frame errors during transmission,frame order and scrambled data are embedded into frame together,so as to recover the inter-frame error,such as disorder,frame loss and so on.At the decoder side,marked data is extracted to reconstruct motion residuals.Combining with the set theory,errors in decoded video sequence are located,and the reconstructed motion residuals are used to conceal the errors.This innovation is mainly reflected in the implement of compressive sensing with high compressing and reconstructing performance.3.An intra-frame error concealment scheme using 3D reversible data hiding(3DRDH)is proposed.Accurate error concealment scheme requires valuable comprehensive information about video,which means that the effect of error concealment is proportional to how much data can be used to recover the error.However,too much marked data will seriously degrade the quality of cover video.Based on these problems,a 3DRDH scheme based on histogram shifting with minimized distortion is proposed.Compared with the existing 3DRDH,the proposed 3DRDH scheme has better embedding capacity-distortion performance.Because of the data size and statistical characteristic,the motion vector is regarded as marked data.In order to resist random channel errors and improve the robustness of error concealment,the duplication of motion vectors in a frame are scrambled together with the original motion vectors,and then all of them are embedded in the QDCT coefficients using proposed3 DRDH scheme.At the decoder side,when the marked data is extracted,the original QDCT coefficients are recovered losslessly.The extracted motion vectors are used to locate the matching block in the reference frame and conceal errors.