A shadow elimination algorithm for vehicle detection by merging projection features and HSV color space

摘要

This paper presents a novel algorithm based on projector features to eliminate dynamic shadow. Through the projected image, this method will split the vehicle, and extract the single-vehicle image. Then using the gradient projection separates regions of shadow and vehicle. Effect of shadow elimination of the algorithm is better, and separated image of vehicles has not holes. In addition, this paper improves the algorithm of shadow elimination based HSV color space by optimizing the Ostu algorithm. Using merging algorithm integrates these two algorithms to further improve the effect of shadow elimination. And performance of the algorithm is more accurate description by using performance parameters of F.