Mean shift based target tracking robust to illuminance variation in infrared image sequences

Abstract

In this paper, a Mean Shift based target tracking approach for infrared image sequences is proposed. Well known Mean Shift method tracks a target in the image using its color histogram. But this approach can fail in infrared image sequences due to several reasons like illuminance variation. However it is possible to improve the performance of Mean Shift approach by using different features. In this paper a feature that does not depend much on illuminance variations is used. Performance of this approach is compared against the standard Mean Shift approach using several infrared image sequences.