A Novel Methodology for Ball Detection and Trajectory Tracking Using Frame Difference and Mean Shift Algorithm

G. Simi Margarat and S. Sivasubramanian

This paper presents the detection and tracking of basketball. The vision based computing received wider acceptance in gaming. The problem of ball detection and tracking is addressed in this paper. For ball detection, frame differencing method and color based object tracking methodology is investigated and implemented. The proposed method uses the frame difference method to detect the ball. A modified mean shift algorithm for tracking is proposed which uses multiple templates to detect and track the basketball. The improvement in precision for the proposed method is about 7% higher than the existing method. The tracking is also improved by about 50% during miss detection and 7% in make detection.

Volume 11 | Issue 7

Pages: 158-171