With network environment improved, video has been gaining popularity, images and text cannot meet the social needs of users, ordinary people are increasingly willing to express emotion by video. Quickly and effectively producing a personalized video become an urgent needs for ordinary people. Based on the analysis of video special effect editing software commonly used, analyze the limitations of ordinary people using video effect editing software, and do some research on user requirement, we propose two automated filter algorithms and a tracking approach based on sparse optical flow combined with SIFT, details are as follows:1. We propose two automated filter algorithms, automated skin beautification algorithm and automated image cartoon algorithm, this automated approach allows ordinary users to get rid of the shackles of expertise and freed from the complex process flow. In automatic skin beautification algorithm, we carry out contrast experiment and choose skin detection in YCbCr color space, parameters of the beautification was determined by the experiment and investigation statistics, result is obvious, the algorithm can be applied to the actual skin beautification. In automated cartoon generation algorithm, we analyze the problems existing in current automated cartoon generation algorithms and the characteristics of cartoon image, a new cartoon automated generation algorithm is designed, the algorithm consists of image edge operation and image smooth quantization, the algorithm meets the basic needs of cartoon generation.2. In tracking algorithm, based on analyses of feature extraction and matching algorithms, we propose a tracking approach based on sparse optical flow combined with SIFT. Experiment shows that the proposed method maintains advantages of SIFT descriptor that is strong, and also improve the speed of feature points tracking.3. With above and other algorithms, this paper develops a special effects system based on the MVC pattern used for ordinary people, which can meet the needs of ordinary people. |