An Improved Mean Shift Algorithm Based Tracking System for Soccer Game Analysis

Abstract — In this study, an improved mean shift algorithm based tracking system with motion prediction and object window enlarging for soccer game analysis is proposed. The objective of the proposed system is to track soccer players (moving objects) efficiently and accurately. Discriminative color selection for modeling soccer players is employed, occlusions are well handled, and incoming and outgoing players are detected and handled in the proposed system. Based on the experimental results obtained in this study, the tracking results as well as the recall and precision rates of the proposed system are better than those of the three comparison systems. I