Computer Vision based Automatic Scoring of shooting targets

Abstract

Computer vision based scoring system can break the monopoly of other automatic scoring systems like shell shockwave amplitude system due to its ease of implementation and cost effectiveness. This paper presents a computer vision based automatic scoring method for the shooting targets. We perform morphological processing of the target image to thicken the boundaries of the bullet hits and then segment the target area by hysteresis thresholding. The impact of illumination variations is handled by adjustable thresholds. The bulls eye of the target is segmented by the help of distance transform to calculate the score inside the bulls eye. Thus, our method is capable of scoring inside and outside the bulls eye separately. The bullet hits are labeled after the segmentation of the target area and the overlapping bullets are also scored by defining a threshold pixel area for the bullet hits. The proposed algorithm is tested on 100 target images with varying number of bullets hit, resulting in bullet count accuracy of 98.3%.