Background estimation and player detection in badminton video clips using histogram of pixel values along temporal dimension
Yahui Peng, Xiao Ma, Xinyu Gao, Fangxu Zhou · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2015
Computer vision is an important tool for sports video processing. However, its application in badminton match analysis is very limited. In this study, we proposed a straightforward but robust histogram-based background estimation and player detection methods for badminton video clips, and compared the results with the naive averaging method and the mixture of Gaussians methods, respectively. The proposed method yielded better background estimation results than the naive averaging method and more accurate player detection results than the mixture of Gaussians player detection method. The preliminary results indicated that the proposed histogram-based method could estimate the background and extract the players accurately. We conclude that the proposed method can be used for badminton player tracking and further studies are warranted for automated match analysis.