Toward a more practical offside line detection method in broadcast soccer video

Lines detection on playfield is an important step in high level semantic analysis of the soccer video. Most of the field lines that stated by the match rule of FIFA can be detected easily. Different from these lines, offside lines are difficult to detect due to its low contrast. To get the binarized regions of these offside lines, the image should be enhanced first. Then, they can be binarized and detected using hough transform. However, it is still not practical to detect these offside lines as the enhanced image is binarized manually in existing method. To overcome this drawback, we propose a more practical method to detect offside lines. In our method, the binarized image is obtained automatically. Although there are false binarized regions and false detected candidate lines, the post-processing steps in our method remove these errors effectively. The proposed method is evaluated over real broadcast video sequences, and experiment results have verified the effectiveness of the proposed method.