Automated Classification of Cricket Pitch Frames in Cricket Video

The automated detection of the cricket pitch in a video recording of a cricket match is a fundamental step in content-based indexing and summarization of cricket videos. In this paper, we propose visualcontent based algorithms to automate the extraction of video frames with the cricket pitch in focus. As a preprocessing step, we first select a subset of frames with a view of the cricket field, of which the cricket pitch forms a part. This filtering process reduces the search space by eliminating frames that contain a view of the audience, close-up shots of specific players, advertisements, etc. The subset of frames containing the cricket field is then subject to statistical modeling of the grayscale (brightness) histogram (SMoG). Since SMoG does not utilize color or domain-specific information such as the region in the frame where the pitch is expected to be located, we propose an alternative algorithm: component quantization based region of interest extraction (CQRE) for the extraction of pitch frames. Experimental results demonstrate that, regardless of the quality of the input, successive application of the two methods outperforms either one applied exclusively. The SMoG-CQRE combination for pitch frame classification yields an average accuracy of 98.6% in the best case (a high resolution video with good contrast) and an average accuracy of 87.9% in the worst case (a low resolution video with poor contrast). Since, the extraction of pitch frames forms the first step in analyzing the important events in a match, we also present a post-processing step, viz., an algorithm to detect players in the extracted pitch frames.