Star Pattern Recognition Algorithm Based On CCD Star Sensor

CCD star sensor is one of the most precise sensors for measuring spacecraftattitude. with low quality and power consumption, it is applied more and more widely.And it was the focus of attitude sensor research. Recognition algorithm is the keytechnology for star sensor, because the real time and correction performance isdetermined by the algorithm. This dissertation focuses on the star pattern recognitionmethod, including that attitude determination method based on the CCD star sensor,guide catalogue building, the estimation of star cancroids on sub pixel, star mapssimulating, and recognition algorithm. The mainly research results include the followingaspects:The mathematical model is built based on the connected-componentsextraction algorithm, and a method of the cancroids localization errorcompensation is proposed for Star target detection.1)Study on target segmentationmethod which based on the connected-components extraction algorithm,throughtheoretical analysis to solve the problem that how to select the threshold of targetseparation,we also solve the problem that how to separate of noise and invalid targetfrom effective target, the experiments show that this method is very effective.2)Thisarticle focus on the method of object extraction, set up the error analysis model ofobject extraction, the expression of the cancroids localization error compensation isgotten, on the basis of this method a compensation method was proposed, the simulationresults show that this method can make the positioning accuracy up to0.005pixels.Star pattern recognition algorithm based on main star pair was set up whichis full of high reliability, the research background of the study was the application ofonboard computer system,the study focus on high reliability identification method,thismethod overcomes the shortcomings of low recognition rate when there are brightnesssimilar goals in the star map,the method can improve the reliability of algorithm by theangle matching approach,it also can improves the accuracy of identifying each time,simulation results show that in the5interference and2pixel error,the method still havemore than97%success rate of recognition.A high reliability method of star pattern recognition algorithm wasproposed,the research background of the study was the ground system, a method whichbased on triangle eigen vector was set up and the noise immunity of this method wasvery strong,this star pattern recognition algorithm through multiple contrast improvesthe interference immunity, the simulation results show that the method has largecapacity guide star catalog conditions, after a long time operation can achieve100%success rate of recognition.The impact of recognition algorithm performance was studied,and one error match probability calculation model was set up.1) The angular distance algorithmmodel is established, and we find an effective algorithm of calculating the observed starnumber.2) On the influence of star map identification rate factors were analyzed and thecorresponding measures was studied.This paper based on the domestic and foreign experience, which has set up targetextraction and separation model, put forward two kinds of optimization algorithm, onewas high reliability and another was full of strong anti interference,it has a certaintheoretical and practical significance, especially in the aspects of target extraction andstar pattern recognition algorithm.