Enhanced Homography-Based Sports Image Components Analysis System

Sports data analysis is an active research area in recent years to automatically extract all the significant information from the sports data. One of the significant tasks is to trace the 3D calibrated sports image to a 2D soccer reference field and map the player’s position in the reference plane. This paper proposed an automated, efficient, and more accurate system for field registration on a reference plane, player’s localization, and team recognition. A feedback-based enhanced homography transformation is used for field registration. Also, the player’s team and the referee are recognized based on their jersey color and marked on the reference plane with their respective team color. The various algorithms constituting vision-based features such as color, shape, size are used to implement the proposed system. Extensive experiments are conducted on different datasets to demonstrate the effectiveness and efficiency of the system in terms of achieving the accuracy to determine the ground truth values of the field, players, and their corresponding team. Additionally, the proposed system’s (comprising enhanced homography) results are juxtaposed with normal homography-based system to outline the improvement.