An improved similarity measure based IR target tracking algorithm

Based on the characteristics of IR target imaging, a simple to implement, good robustness target tracking algorithm is proposed, which is the result of a theoretical analysis of the similarity measure function of the core component of the original algorithm. Two improvements are given. First, based on the idea of clustering, dynamic weighted coefficients are added to the matching samples in similarity measure function, which, in effect, improves the matching accuracy of model image and target image. Second, the pixel neighborhood information of model image and target image is augmented to the original model, which, in turn, further improves the tracking algorithm's robustness. Experimental results show that the new algorithm can achieve good result of tracking IR targets under complex scenery, and this demonstrates the feasibility and effectiveness of the algorithm.