A novel error concealment approach based on general regression neural network
Shih-Chun Shao, Jun-Horng Chen · 2011
In video communication, packet is inevitably lost or transmitted erroneously over error-prone channel. If there are packets lost, the entire video quality will be degraded. The error concealment is thus proposed to solve this problem effectively. Therefore, this paper will propose the general regression neural network (GRNN) can be used to estimate the motion vectors of the corrupted macroblocks. The proposed approach can restore corrupted frame effectively. Experimental results show that the proposed approach in this work can improve the defect of conventional approach and raise the average PSNR of the recovered video sequence.