| The sports video occupies the important proportion in resources of the video, so how to classify, organize and retrieve it has become an important research field. Traditional video retrieval can only search manually through soon entering and soon retreating method, so it is a very tedious work that consuming time. This paper uses soccer video as an example, carries on content-based video semantic analysis, and realizes the goal shots detection. The purpose is to bring the convenience for the video retrieval and video automatic editing.This paper has carried on the following research mainly: first, shot segmentation is a primary work of content-based video parsing. On the basis of drawing experiences from the existing shot segmentation methods, this paper combines the special characteristics of the soccer video to propose a double threshold pixel-matching twi-difference algorithm. It achieves a comparatively satisfied result of segmentation. Second, using the characteristics of goal shot, this paper purpose a goal shot detecting algorithm, namely domain color area and projection detection. The complexity of the algorithm is low, and the accuracy is high. It meets the request for the real-time detection. Third, this paper has explored the video highlights generation and the video automatic editing. The fourth, it offers four-level retrieval structure of "key frame - shot - highlights - video", and can play shots or highlights.The content-based soccer video retrieval and editing system, which the arrangement is clearly demarcated and the interface is handled friendly, can successfully detect the goal shot and satisfy the user's real-time retrieval. This paper goes on a large number of experiments, and also points out the later research direction.This paper realizes the purpose of automatic retrieval and editing according to goal shot. Though there is distance with practical stage, it provides a good technological foundation for further research.
|
PDF Full Text Request