Detection of Multi-target Based on Maximum Confidence

A multiple targets detection algorithm based on maximum confidence is proposed to solve complicated multiple targets track initiation problem in the mode of hybrid motion with insufficient prior information, such as target number and target velocity interval. This algorithm effectively utilizes the signal intensity information of sensor depending on the principal of energy accumulation of dynamic programming. During the period of rough initiation, an extended search approach is utilized to generate candidate tracks, and these tracks are divided into rectilinear motion mode and curvilinear motion mode respectively according to model matching. Furthermore, the probabilistic multi-hypothesis tracking algorithm based on signal intensity information is utilized to obtain the maximum confidence through calculating optimal state estimation, and the target measurement is confirmed finally based on the maximum confidence in the period of track confirming. Simulation results show that the proposed algorithm has the capabilities of high real-time and accurate initiation for multi-target tracks. Moreover, it avoids the error tracking problem due to poor initial value in probabilistic multi-hypothesis tracking algorithm.