Improved infrared target-tracking algorithm based on mean shift
Zhile Wang, Qingyu Hou, Hao Ling · Applied Optics · 2012
An improved IR target-tracking algorithm based on mean shift is proposed herein, which combines the mean-shift-based gradient-matched searching strategy with a feature-classification-based tracking algorithm. An improved target representation model is constructed by considering the likelihood ratio of the gray-level features of the target and local background as a weighted value of the original kernel histogram of the target region. An expression for the mean-shift vector in this model is derived, and a criterion for updating the model is presented. Experimental results show that the algorithm improves the shift weight of the target pixel gray level and suppresses background disturbance.