As the existing spatial error concealment algorithms is lack of the recovery accuracy in complex texture areas

a spatial error concealment method based on tensor voting was proposed

which extracted structural information by using tensor voting.On this basis

an improved spatial error concealment algorithm which combined with the traditional algo-rithms was further proposed.Firstly the lost marc blocks were classified into three categories

and then the appropriate concealment method was chose to recover the lost macro blocks according to the type of lost macro blocks.Experimental results show that the proposed method improves the recovery accuracy effectively and avoids the edge disorder phe-nomenon

especially in the region with more complex texture

the subjective and objective quality of concealed image is improved greatly.