The judgment of sport games is a subjective task, especially for some skillful competitions. The subjectivity brings many unfair judgements to the competition. It can be used to effectively decrease the unfair judgements by using computer to judge the competition automatically or assist the referee to judge. Synchronized diving is a very skillful competition and the whole action has to be completed in an instant, which makes the judgement very difficult. In this thesis, a framework is developed to analyze and evaluate the synchronization in synchronized diving. The framework is composed of three steps: (1) Moving object extraction based on dynamic background construction; (2) Feature representation and extraction for synchronization; (3) Synchronization ranking. To the best of our knowledge, it is the first time that the idea of ranking which is a common method in information retrieval is introduced into the synchronization evaluation problem. The absolute scoring problem is converted into the relative ranking problem, and finally the problem is solved. The main contributions of this thesis include the following issues: (1) The athlete extraction is designed and implemented based on dynamic background construction in synchronized diving video. We use a motion detection method with consideration of the global motion estimation, to exactly extract athlete from the video. (2) An efficient method is presented for synchronized feature representation and extraction according to the synchronization judge rules. The extracted features can describe the synchronization very well by bridging the low-level visual features and the high-level judgement rules. (3) A synchronization ranking method for synchronized diving is proposed based on synchronization judgement characteristic by using preference learning approach. The ranking method avoids the usage of the absolute numeric synchronization scores, which reduces the difference existing in the different situations and the different referees.