Novel adaptive intra-frame error concealment algorithm for satellite video

Bit error and packet loss exist during the transmission of video in the error-prone satellite channel. According to the edge feature in local video, each lost block is categorized as flat block, edge block, or texture block. Then bilinear interpolation, directional interpolation or novel directional weighted interpolation is adaptively selected as the interpolation method to improve reconstructed video quality and speed up the concealment process. Noise pixels are eliminated from the candidate prediction list to reduce the gradient computation. Prewitt mask is chosen as the gradient filter to accurately present the image directional information and restrain noise. Moreover, fractional pixel interpolation scheme is used to increase the interpolation precision and adapt the application of high-definition video. Objective evaluation shows that a peak signal to noise ratio(PSNR) improvement of up to about 3 dB with negligible decoding time increment can be obtained. Subjective comparison shows that our proposed algorithm can achieve higher visual quality.