An effective algorithm of tracking multiple objects in occlusion scenes

In the field of Computer Vision (CV), multiple object tracking has been one of the hot issues, and it is especially one of difficulties to effectively track multiple objects in occlusion scenes. In this paper, we propose an effective algorithm to dissolve the occlusion issue in tracking multiple objects, combining the feature of overlap region with the feature of RGB-based dominant color histogram (DCH). Firstly, occlusion can be judged by computing the size of overlap region. Secondly, if occlusion takes place, RGB-based DCH can be used to estimate the likelihood of occluded objects. And then, the occluded objects are tracked effectively. Additionally, RGB-based DCH is only computed when merging and splitting, and the computational complexity is effectively reduced. From the experimental results, the proposed algorithm can improve the validity and robustness of tracking multiple objects.