Research And Implementation Of Soccer Video Retrieval Technology

Currently, multi-media have been used in many fields, such as sports, journalism and other fields, even family shoot. And the development of multi-media technology also makes its usage more widely. Widely usage brought about a large number of video data. How to manage these data efficiently is an urgent problem to solve. But especial structure of video data is another difficulty for solving this problem, so this field is a hotspot in science research which makes many researchers at home and abroad participate in this project.Scholars define research which finds ways to solve the managing of video data efficiently as Content-based Video Retrieval. In the beginning of this paper the development and current situation of Content-based Video Retrieval technology had been introduced, and then present the principal work this paper does. Actuality and achievement of Soccer Video analysis was recommended in this paper also. And it gave the presentation of shortage in Soccer Video analysis.Principal works have been done in this paper have three points. Firstly, shot Boundary Detection Based on Incremental Principal Component Analysis was proposed to detect the boundary of video shot. Its advantage is detecting shot boundary and extracting abstract without acquiring the whole video in advance, in virtue of which it make the online detection come true and reduce the dimension of the feature space. The essence of this method is to compute the similarity between frame to the frame' principal character. Secondly, a modified way to extract the abstract of video which depends on the rank of a slide window was referred in this paper. The slide window composes of video frames. This method has the characteristics of less calculation, high velocity, high efficiency and etc. Finally, the way to extract feature space of soccer video which considers the specific characters of soccer video structure was introduced in this paper. Then describe how to detect the highlight of soccer video on the basis of feature extracting.The system for detecting goal event was shown in the last of this paper. Method based rules was used to detect the goal event which is convenient to some extent. Its' key point is how to build the rules collection. It affects the result of detection. Computing the similarity of the key frames between sources and example helps accomplish the retrieval of video.