Digital Video Post-processing Algorithms

As technological progress and development of the Digital TV industry, there are more and more videos in different formats which are transmitted through different networks and played on different terminals. In order to get good results of videos in different formats played on different terminals as far as possible, the digital video post-processing becomes more and more important.Digital video post-processing algorithms mainly cover video format conversion and image enhancement. Video format conversion algorithms mainly cover de-interlacing, sample-rate conversion and frame-rate conversion. Image enhancement algorithms mainly cover digital noise reduction, lightness transient enhanced and gamma correction. The key algorithms of digital video post-processing are discussed in this paper, and the optimization and realization of de-interlacing and image interpolation are also presented in this paper.In regard to the de-interlacing algorithm, the basic principle and effects of de-interlacing algorithm are reviewed first, then all kinds of conventional algorithms concerned by de-interlacing and their performances are discussed. After that, a motion compensated de-interlacing algorithm based on parity property of fields is presented. Motion vectors are estimated from bi-directional motion estimation; the parity property of fields is used in the error control of motion vectors and interpolation process in order to improve the accuracy of the search and interpolation, and the transmission errors are avoided; the spatial correlation of motion vectors is used to generated a adaptive threshold, the compensation interpolation and the intra-frame interpolation can be chosen in the de-interlacing process. Finally, the simulation results comparing with conventional de-interlacing algorithms are given.In regard to the image interpolation algorithm, this paper first reviews the process of ideal interpolation, then all kinds of conventional algorithms concerned by image interpolation and their performances are discussed. After that, an image interpolation algorithm based on the edge and texture of videos is presented. The pixels to be interpolated are dived into flat, edge and texture categories, different interpolation methods are applied for pixels of different categories after the division. Finally, the simulation results comparing with conventional image interpolation algorithms are given.The de-interlacing algorithm and image interpolation algorithm presented in this paper get better results of subjective and objective standards, but the complexity of them needs to be improved, and they also need to be further optimized for the hardware implementation.