A computer vision system for chess game tracking

In this study, a real time computer vision system that can detect and analyse moves in a chess game to support referees' decision making processes is proposed. A setup is established for suggested system to detect moves during a chess game. In this setup, there is laptop computer and a RGB web camera, which is positioned over the chess board. All of the steps in the process are programmed in MATLAB computing environment. The proposed computer vision system is not using any third party chess engine, and it is handled and tested in two parts; (i) move detection and (ii) move interpretation. In move detection part, snapshots taken by web camera are processed one by one. For each snapshot, color and position of each chess piece are computed. In the preprocessing stage, outside border of chess board is calculated and the image is geometrically adjusted and enhanced. In move interpretation part, color changes over the chess board are analyzed and the system decides whether these changes are interpreted as a move by player or not. If it is a move by a player, this move is validated whether it is a legal move or not. Results are obtained and user is notified accordingly. Besides, the results are visualized in a 2D environment and logged by the proposed system. Tests show that the system is evaluated to be successful in both move detection and interpretation under different illumination conditions.