Robust Object Tracking in Infrared Video via Adaptive Weighted Patches

With the quick development of computer and electronic techniques, infrared sensor-based object tracking has become a hot research topic in recent years. However, infrared object tracking is still a challenging task due to low resolution, lack of representing information, and occlusion. In this work, we present an adaptive weighted patch-based infrared object tracking scheme. First, the candidate local region is divided into non-overlapping sub regions, and a set of belief weights is set on these patches. After this, a particle filtering-based infrared object tracking system is realized. In the last, the belief weight of each patch is evaluated based on the linear discriminative analysis (LDA) and particle sampling scheme. Experimental results on challenging infrared sequences show that the proposed algorithm can effectively locate the tracking object.