Research On Key Issues Of Analysis Of Basketball Video

With the development of multimedia technology and the appearance of information highway, the storage and transmission of digital video has become a less difficult issue. Various sport games bring the explosive expanding of sport video, therefore, efficient analysis of sport video is necessary.In this paper, three key issues are studied in the analysis of basketball video: shot segmentation, shot coarse classification and detection and location of scoreboard. The following algorithms are proposed:1. An algorithm of shot segmentation is proposed. Firstly, the data in each frame is quantized in HSV model; then based on histogram of each frame, the difference of adjacent frames, the difference of cumulate frames and the percentage of the frame difference in a defined window are computed; furthermore, the adaptive average in a window, the adaptive thresholds and sliding windows are obtained. Finally, shot segmentation is implemented using the above parameters.2. A algorithm of shot coarse classification is proposed, firstly, key frames are extracted in equal interval in a shot; then for a frame, Dominant Colors are obtained by Hue, furthermore, the variance of Dominant Colors and Ratio of clustered colors are computed, by which the corresponding frame could be classified into playing or non-playing frame, Finally, a shot could be classified by the percentage of playing frames.3. An algorithm of detection and location for scoreboard is proposed. Firstly, the template of scoreboard is built using multi-dimensional correlation, it involves the location and the detection parameter. Then the scoreboard is located by the template matching.