A new snake model robust on overlap and bias problems in tracking a moving target

Abstract

Performance of conventional snake models tends to decrease in the case where the target object is overlapped with other objects and where the bias between the initial snake and the target object is large. To solve these problems, this paper proposes an extended snake model including the energy dissipation function in the external energy, which is defined as a function of the variation rate of the target object' area and the target object's velocity obtained by the modified SSD algorithm. In the experiment, it is shown that the proposed snake model renders the mobile robot with a camera follow the moving target successfully even when it is occluded or overlapped temporarily by other objects.