Mean-Shift Algorithm Apply for Infrared Imaging Tracking

An improved algorithm based on the modified histogram & mean shift is proposed in order to deal with difficulties in tracking small target of infrared imaging under complex background. The new algorithm is based on a lower dimension combining characteristic weighted histogram, the analogical degree measurement is determined by the Bhattacharyya coefficient, and the target is located through the mean shift. In order to improve the precision and robustness of the modified mean shift algorithm, particle filtering is used to obtain the optimal position estimate. Test results show that, the tracking point fluctuation could be reduced to no more than 2 pixels with the modified algorithm, for the test case.