Study On The Space Object’s Attitude Measurement Based On Multi-linear CCD System And Multi-point Cooperation Target

With the constant development of science and technology, the attitudemeasurement of space object is used more and more widely in different fields suchas aviation, aerospace, navigation and military and so on, in which differentmeasurement methods are proposed. When measuring the spatial moving object’sattitude in the suspension state, the common methods are unfit for measuringattitude that is fast-changing, because of the disadvantage of system complexity,high cost, low resolution and slow speed of image sensor and so on. Therefore, it isnecessary to research a new kind of exterior attitude measurement system forattitude which is fast-changing in the suspension state.The objective of the subject “Study on the space object’ attitude measurementbased on multi-linear CCD system and multi-point cooperation target” is to researchthe attitude of spatial object that is fast-changing in the suspension state, develop asimpler exterior attitude measurement system which improves system performancefor solving the problems of traditional exterior attitude measurement system, buildan effective attitude calculation model, discuss how to improve the linear CCDcalibration precision, and provide the theoretical basis for high-precision attitudemeasurement. The research of this subject can be applied to the suspensionexperiment of spatial object’s attitude measurement, calibrate the internal attitudemeasurement result based on inertial navigation device installed in the space objectinternally, and its application prospective in measuring three-dimensional motionsis very vast.The basic principle of exterior attitude measurement system that based onlinear CCD and the traditional exterior attitude measurement system are described.The traditional system is too complex due to the nine linear CCDs and six camerasit adopts, and the constraints of these camera positions may cause many problems,such as non-linear system error, poor uniformity of measurement and morecalibration parameters. In order to overcome these problems, a new system whichuses only3linear CCD cameras was proposed, and this system can capture multipleassistant points that are illumed simultaneously. The1D camera is composed of alinear CCD and a cylindrical lens, the affect of arrangement of these linear CCD cameras to space measurement was analyzed, and which can be used as theoryinstruction of the experiments.In order to improve the speed and precision of attitude parameters calculationand real-time performance of system, many attitude representation methods areanalyzed. According to system’s structure and characteristics, a model is presentedbased on Rodrigues parameters, which is used as the attitude representationparameter, combined with relationship of the intersection vector between linesegments which are composed with point cooperation targets. At the same time,four groups of parameters whose norm would not be infinite simultaneously switchto avoid the singularity. The attitude calculation model based on Rodriugesparameter to calculate the attitude for the space object is proposed. Meanwhile, theinfluence of spatial point’s position error is analyzed for the attitude calculationmethod, and the sensitivity to spatial point’s position error of different attitudecalculation method is researched.In order to improve the calibration precision and speed of linear CCD camerameasurement system, a method of linear CCD camera calibration based on BPneural network is proposed by using the ability of neural network which haseffectively dealing with complicated nonlinear mapping relationship from input tooutput. By training the nonlinear mapping relationship between spatialidentification point on space object under3D coordinate and pixels under imagecoordinate system, this method solves the problem that the nonlinear errors causedby camera lens distortion and system errors in the calibration method of DLT, andachieves the linear CCD camera calibration of the exterior attitude measurementsystem and the reconstruction of3D information of spatial cooperation target.In the last section of the paper, the method and model proposed in this subjecthave been verified by experiments. Firstly, the linear CCD camera calibrationmethod based on BP neural network has been carried out through experimentalstudy. Experimental results indicate that, the error fluctuations of BP neuralnetwork method in calibrating linear CCD camera is decreased at93%than DLTmethod. Then, the attitude calculation model of exterior attitude measurementsystem has been carried out through experimental study. Experimental resultsindicate that, compared with the quaternion method on the premise of assuringprecision, computational speed is increased at37.6%. In the actual measurement, it illustrates that the algorithm is superior to the quaternion method.