Research On The Detection And Tracking Of Billiards In Snooker Video Based On Kalman Filter

In this paper, detection and tracking of moving billiards in the snooker competition videoare studied. Detection includes background extraction, billiard segmentation and objectpretreatment; In the tracking the method of Kalman filter and target matching and the methodof Kalman filter and mean shift are used. Follows:Target detection, first introduce the method commonly used for target detection andbackground extraction method, and several background extraction are compared and analyzed;background subtraction method is used to extract movement billiards targets and extract targetpretreatment.The moving target tracking combining Kalman filter and target feature matching methodto track motion billiards. First introduce the common methods of moving target tracking, aswell as Kalman filtering theory; then the billiards centroid features, the area features and colorfeature are combined to match the target; Finally, prediction of the Kalman filter andcharacteristics matching is used to track target. Experiments show that the algorithm isfeasible and effective in the absence of shelter conditions.For tracking under occlusion, according to the characteristics of billiards color, thecombining method of Kalman filter and the Mean Shift is used. First introduce the Mean Shifttheory and realization of Mean Shift algorithm in tracking; then prediction of the Kalmanfilter and Mean Shift is used to track target. Experiments show that the algorithm is feasibleand effective in blocking situations.