Image Post-processing Main Algorithm Research And Realization

Video sequence post-processing is a very important part of the whole video sequence processing. It's a bridge between the signal channel and the terminal device. With the tremendous developing of video sequence processing and the huge emergence of novel video sequence applications, video sequence post-processing is carrying more meaning and requiring higher speed and better effect. Video sequence post-processing is, nowadays, not only composed of the traditional algorithms such as noise reduction, de-interlacing, scaling and enhancement; but also embodied with increasing real-time applications such as image tracking and face recognition.This paper is based on the research of video sequence post-processing's key algorithms. The background of the research is first introduced and then the main intents of video sequence post-processing domain and the research status from home and abroad. After that, the paper focuses on the de-interlace domain and the image tracking domain. A novel de-interlace algorithm based on the enhanced edge self-adaption is presented. On one hand, this novel algorithm optimizes the edge self-adaption method based on the research of the leading de-interlace algorithm; on the other hand, the fuzzy function is introduced to make the recovered video sequence more smoothly. Beside the de-interlace algorithm, a new image tracking algorithm based on the weight-parallelism particle filter is brought forward. First, the dependency of the traditional particle filter algorithm is broken for parallelism. Second, a multi-processor architecture is presented to elevate the processing speed. By experiments, the effectiveness of these new algorithms is proved. And the hardware realization of these two algorithms is presented. The last part of this paper concludes our work and gives an outlook of the video sequence post-processing domain.