Point Target Detection And Recognition Algorithms In Optical Image Terminal Homing System

This dissertation addresses the problem of point target detection and recognition algorithms in optical terminal guidance system, which is the key technology for KKV intercepting the ballistic target far away. The Detection Algorithms for low SNR point target, the Ballistic Target Signature Analysis, the Dynamical Target Recognition Algorithms on Seeker are systematically studied in this thesis.The detection algorithm for the low SNR point target in image sequences and its performance is studied. Based on the continuity of target's trajectory, a detection algorithm formed by modified DPA and track association algorithm is presented. The modified DPA energy integration algorithm with a confined search field is developed to solve the target energy pervasion problem. The gained SNR is higher than the previous algorithm's. A Linear Least Square Predictor is presented for the trajectory association algorithm. It can effectively reduce the prediction error and improve the trajectory detection probability. With the result of theory analysis and simulation experiments, it can be shown that the detection goal can be accomplished by the developed detection algorithm, and at the same time the algorithm has the structure of easily implemented in hardware. A parallel distributed sequential fusion detection algorithm is also developed for the multi-band target detection problem. The optimized fusion algorithm is determined based on the theory analysis and the experiment's results. The exo-atmospheric ballistic target's optical signature and recognition feature is studied. The exo-atmospheric combat enviroment is introduced, including the outer-space's optical signature, the optical and dynamic signature of the potential targets. The research focus is the precessing RV and tumbling decoy's irradiance intensity's periodical changing rule. A double periodical function model is presented to simulate their irradiance functions. The method of calculating the equivalent temperature using double IR band radiant intensities is also studied. From the RV and decoy's equivalent temperature data sequences, the temperature feature is extracted to discriminate the RV and decoy, and the decision rule is designed.Dynamic point target recognition system and discrimination algorithms on seeker board are studied. The dynamic recognition process through the intercepting course in out-space is analyzed, and the method of dynamic managing the target's record data and dynamic recognition information processing model are developed. A dynamic recognition algorithm employing the FIR-TDNN classifier for RV and decoy's discrimination is developed. The algorithm analyzes their temporal radiant intensity sequences, and assesses the lethality of the potential target and judges which target is the most lethal one. With the theory analysis and simulation experiment results, it can be shown that the algorithm can effectively solve the lethality assessment problem.