A Novel Frame Error Concealment Algorithm Based on Dynamic Texture Synthesis

The traditional frame error concealment methods primarily predict the motion trend for the corrupted pixel or block, which would generate edge fragmentation and object deformation. When the damaged images contain the contents of the non-linear motion and the global illumination change, these methods would have bad effects. In order to improve the efficiency of frame error concealment, we propose an improved frame error concealment method using dynamic texture extrapolation. First, we give an improved solution of the dynamic texture model, which is good for video coding.