Based Video Pre-processing, Fuzzy Theory

The video pre-filter, a technology of video enhancement, tends to be popular due to the video standard and computer hardware advance. A fuzzy video pre-filter method is proposed based on in-depth research and analysis. Furthermore, the new method coupled with the H.264 video coder presents a fine corporation by communicating some variables between these two machines.The fuzzy video pre-filter method (FVPFM), effectively reducing impulse or mixed noise, encompasses a detection phase and a filtering stage. The detection section is a method to obtain high hit-rate of coverage and correct rate by calculating fuzzy gradients. The noise models tend to be estimated by the first detection with a high correct rate. The new filter processes the noise pixel by employing membership functions and weighing the relationship of the adjacent pixels. The filter, specially designed for video processing, includes intra-filter method and inter-filter method. The experimental result presents better value of correct rate, coverage, PSNR and video quality compared with the classic spatial filters.The proposed method entwined with the H.264 video coder is beneficial to the speed and accuracy in the video processing by communicating the key parameters. The new approach has an assortment of variables, some of which is related to the speed of processing and others to the accuracy, owing to the implement of fuzzy mathematics. Meanwhile, bulk of parameters of the inter-frame search algorithms affect the video processing in the coder. The feedback from the coder to the pre-filter is a simple and good resolution to match the similar speed between two machines. The result of using the new method together with the coder shows less codec bits, less codec time-cost and better video quality compared with the single coder when the video sequence is corrupted by 5%-density impulse noise.