Morphology based particle filtering algorithm for point target tracking

The problem of tracking maneuvering point target in optical image sequence using a single sensor is considered in this paper. A morphology based method is proposed to pre-detect target area. This pre-detection begins with expansion of the target area by dilation using a selected rectangle structuring element. Then the target energy is enhanced by frame-addition. Finally, the target area is detected by getting the maximum intensity based on the character of target correlation between frames. Therefore, the conventional uniform distribution of particles over the whole sensor field-of-view is shrunk to the pre-detection target area. The proposed morphology based particle filtering algorithm thus enhances the particles representation. To demonstrate the efficiency of the proposed method, it is compared with the particle filter proposed by Salmond. The simulations indicate that desired tracking quality is ensured by narrowing initial particle distribution derived from morphology based pre-detection.