A robust implement of Tabu search based video object tracking

Video object tracking technology has important application in the fields of traffic, military and others. In this paper, a color histogram related video object tracking algorithm is proposed by using Tabu searching theory. First of all, object's features in HSV and RGB color space are extracted, and then we use either entropy or Bhattacharyya to calculate the similarity between raw feature and current feature, which will be used to identify the object in the Tabu search algorithm. After that, we choose the appropriate length of Tabu list and the radius of searching neighborhood. In that case, we reduce the computing complexity under the conditions of ensuring the accuracy, and improve the real-time of tracking. The simulation result shows that the proposed algorithm can adapt to both the static video and the dynamic video well. Besides, it has a better robustness and can achieve the expected effect.