Abstract:

To overcome the inaccurate estimation of motion vectors in temporal error concealment, a motion vector recovery method based on the minimum variance of boundary of motion vectors is presented. Furthermore, a novel temporal error concealment algorithm based on relative strength of motion vectors (RSMV) is proposed, which efficiently exploits the spatial correlation of motion. For a lost block, the RSMV of its neighboring area is first calculated according to the motion vectors of adjacent blocks. Then, either the minimum variance method of boundary of motion vectors or Lagrange interpolation method is employed to recover the lost motion vector according to the magnitude of RSMV. The simulation results show that compared with the traditional algorithms, the proposed algorithm can recover images with high quality and obtain a improvement of about 0.3 dB~3 dB in PSNR in different rates of lost block or different video sequences.