Bat Detection and Tracking toward Batsman Stroke Recognition

This paper adopts a new approach to the problem of cricket stroke recognition from pre-recorded video footage. The proposed method ensures bat detection via Optical Flow and Otsu's Tresholding, thereafter using the Kalman filtered result to train data-sets that can be matched via the cross-correlation function. The bat detection and tracking methods work well, with average positional error rate of approximately 5% with tested footage. The bat tracking method results in valuable positional information that is lacking effective implementation in most current image processing methods.