Research On The Key Technology Of Scene Matching Terminal Guidance System

The precision guided munition with image-aided INS system has high precision and huge efficiency compared to cost, and it will be certain to play an important role in the future war. The use image-aided terminal guidance system can adjust the accumulative INS error, and then lower the requirement of the INS precision. Scene matching is an effective image guidance mode.With the background of terminal guidance using image imformation on low-cost strapdown seeker, this thesis studies deeply in some key technology of scene-matching system, which includes image rectification algorithm, suitable matching region selection, search strategy of matching algorithm and the reliability of matching results.The thesis is mainly concerned with the follows:First, a precise image rectification algorithm is presented for correcting the gross distortion between the real-time image collected by camera and the reference image stored in the missile. Based on the attitude and position information provided by INS, the rectification algorithm can switch the real-time image to the one with the same point of the based image. Experimental results indicate that the algorithm presented is excellent in precision and robustness,and it can improve the precision of matching greatly.Second, the algorithm of suitable matching region selection is presented based on five criterions. Then the paper deeply analyzes the time complexity and pertinency of the five criterions, which can provide guidelines for selecting the suitable region.Third, the search strategy of parallel mountain climbing, and serial mountain climbing with position restriction, and random mountain climbers are studied as well Genetic Algorithm. The paper studies the relationship between the matching probability and the number of original climber that can ensure searching successfully, and it presents some measures on overcoming the slow-moving area and self-adaptive variation stratage.Fourth, the reliability of matching results is discussed. The training samples and the testing samples used to train and test a BP network are created by an emulational platform. The trained network is adopted to measure matching reliability. Then based on the maneuverability law of the carrier in matching process,the paper presents a method for eliminating and reconstructing the mismatching points when the flight trace is approximated to a straight line.The paper studies main aspects in matching system, based on the request of system in real-time, accuracy and stability. The goal of the design is achieved at last.