The Star Map Pretreatment And Recognition Algorithm Of Star Sensor Based On B1000

At present, the star sensor is more and more widely used in calculating for spacecraft, and APS star sensor research has become a hot topic in the field of spacecraft attitude control. CMOS Star sensor is used to collect the star image, which can be used to determine the star sensor’s instant real-time posture after pretreatment and identification for the star map, and then obtain the attitude position of the flight vehicle based on the coordinate transformation.Star map recognition is the premise of stars accurate location obtained, the actual shooting star image by the star sensor, in addition to optical signals generated by the target star point, there are a variety of noise signal. In order to get the exact information of star position and gray, take some preprocessing operations for image data obtained by star sensor, so as to improve the centroid location accuracy, and improve the precision of star sensor output attitude finally. In this paper, the design realized based on images of homemade chip B1000 star sensor star preprocessing and recognition algorithm in analysis on the basis of the existing technology, combined with the needs of the project, design the image pretreatment method and star map automatic recognition of matching method.By analyzing the shooting star picture, noise, take the stripes horizontal stripe filter HVSF algorithm to filter the non-uniformity noise, use the adaptive threshold segmentation and single point of noise exclusion, bad handling, and the background fixed noise processing for further action, obtain a high accuracy star position with belt threshold of qualitative method.Use a automatically recognition method for map recognition with combination of the fully autonomous star identification and local identification: when the Star sensitive device working actually, the system use full days recognition algorithm for attitude recognition to capture attitude if it lose attitude, and take local recognition algorithm to match in remaining time. When match failed, the algorithm will calls fully autonomous star match identification algorithm again, which has a high match success rate and attitude precision with the automatically switch.In addition, in order to verify the algorithm effect, increase the experimental data, a map simulation is used, which can generate the realization of the digital image. On the basis of the traditional research, the new map simulation can get more realistic simulating map images by analyzing the corresponding relationship of radiation intensity of illumination and the gray value, combined with asymmetric two-dimensional Gaussian point spread function(PSF).Algorithm process easy to implement, meet the design requirements for star sensors, is advantageous for the parallel processing hardware, and provides great convenience for embedded hardware implementation. Experimental results show that this map preprocessing technology improves the accuracy of the centroid localization, and the error of star-point localization can reach up to 1/50 pixel after HVSF processing.For ideal star data, star pattern matching success rate can reach 99%, and the rate is more than 98% which meet the attitude detection precision of 0.01 degree. Then after pretreatment, the success rate of the Star map with noise can reach 99%, and the rate is more than 97% which meet the attitude detection precision of 0.01 degree. Algorithm is easy to implement and meet the Star tracker project design requirements, facilitate parallel hardware processing for embedded hardware with great convenience.