Researches On Targets Detecting And Tracking For Image Homing Guidance

Infrared (IR) homing guidance information processing system (IRHGIPS) is one of the most important components of interceptor, the key problem is how to detect and recognize the moving point targets (MPTs) from the background and decoys at long distances, robustly track blob targets in imaging guidance. In this paper, four main aspects of this system are studied:The development of efficient detecting algorithm for targets in low signal-to-noise (SNR) IR images is of critical importance for improving interceptor target detection range in space. According to some properties of space IR images, a two-stage algorithm for the detection of MPTs is proposed. Firstly, an exponential damped and velocity-direction weighted dynamic programming algorithm is employed to accumulate targets signal along their trajectories and obtain candidate targets. Then, targets are recognized using trajectory association and confidence testing. The above approaches have the following characters: high detection probability, simple structure, easy realization by hardware and software and excellent real-time performance.High performance detection algorithm for targets in cloudy image sequence is the key of improving interceptor target detection range in the sky. Light of sight (LOS) stabilization jitter, which results in translational, rotational distortions in registered images, is an obstacle for efficient temporal filtering of frames and target detection. An adaptive spatial-temporal filter is proposed and used to detect and recognize targets. Firstly, an interleaved multimode particle filter (IMM-PF) is presented to accumulate targets random profile temporal signal. Sequentially targets and clutters are recognized according to their difference characters in temporal and spatial luminance. Theoretical analysis and simulations indicate that the algorithm has good performance of detecting moving dim targets in cloud clutter, even when sensor shivers.Accurate,fast and robust tracking maneuvering target is critical important for interceptors to successfully hit the target. The main challenge is to maintain the correct identity to targets when there are disturbance, for example decoys, engine fragments etc., in the sight. The paper proposes a new approach to this problem by incorporation all targets'spatial-temporal context information in target recognizing. The approach is easy realization by hardware and software and excellent real-time performance. The experiments show robust tracking performance.The paper discussed the system hardware structure, designed a real-time operation system (RTOS) based software and hardware realization of the above algorithms. The main work of the paper was used in the IRHGIPS of"863". Test results indicate that the system is reliable and fast.