Video Scan Format Conversion Chip Algorithm

Current video systems employ an interlaced scan scheme with a frame rate of 50HZ or 60HZ where visually annoying artifacts, such as line flickers, large area flickers, and edge flickers, are inevitably generated and more objectionable with a larger display. To solve this problem, current video signals must be converted into deinterlaced signals that have a frame rate of 60HZ. In addition, scaling function is also required to improve display quality. So, scan rate up conversion algorithms must be employed to build a video system (and relevant chip) that will match all the requirements. In this article, first, the theory of sample rate conversion is reviewed. A series of previous algorithms are studied, such as Intra-Field interpolation, Inter-Field interpolation, Motion-Adaptive interpolation and Motion-Compensate interpolation. Their profits and drawbacks are carefully discussed and some experimental results are demonstrated. This article proposed a novel motion estimate algorithm which is based on block-matching algorithm(BMA). The BMA is the most popular algorithm in this area, which is widely used in many applications such as deinterlacing, image coding, etc. However, BMA has its own hold-backs. The size of block is often hard to determine, if a big size is chosen, the so called "blockiness"will cause trouble, if a small size is used, the algorithm often get a erratic motion-field which is not able to express the real motion. In this article, pattern recognition method is introduced to solve the problem, the algorithm use clustering method to smooth the motion-field, so it can choose a small block in BMA. The complication of BMA and clustering algorithm is a big problem in the realization of algorithm. In this article, a hierachical method is applied to optimize the algorithm , and the efficiency of the proposed algorithm is greatly improved. Based on the proposed motion estimate algorithm, this article bring forward a motion compensate deinterlacing algorithm. The main idea of this algorithm is to use the median of motion-estimate based interpolation and traditional intra-field interpolation as the interpolation result. This algorithm can only be realized by software with the assistance of hardware. For pure hardware realization, the author proposed a motion-adaptive deinterlacing algorithm. In this algorithm, motion-detection is performed to determine whether the current pixel is in a motive area or not. If the current pixel is moving then the algorithm use inter-field interpolation, else it...