Research On Optimization Of Star Rattern Recognition Algorithm For Nano Star Tracker

Star sensor is a very important part of the satellite attitude control system.It is widely used in astronomical navigation system.The attitude measurement is carried out with reference to the star as the reference.By detecting the stellar position in the different position of the celestial coordinate system,To determine the attitude of the spacecraft for satellites,spacecraft and other spacecraft and autonomous attitude control system to provide accurate spatial orientation and benchmark.With the satellite miniaturization and autonomous navigation become more mature,star sensor in the absence of posture cumulative error,high precision,strong autonomy and concealment and other advantages gradually highlighted.Star map recognition as the core of the star sensor to achieve the characteristics of the core part of the high accuracy and efficiency of the star map recognition algorithm has been the focus of research at home and abroad related issues.Based on the actual project development,this paper designs and realizes the algorithm of all-day self-contained CMOS micro star sensor.The algorithm includes three parts: star image detection,star image recognition and attitude acquisition.Among them,the core part of the star map recognition algorithm in the software and hardware aspects of a comprehensive optimization.In order to overcome the redundant matching caused by the low feature feature of the triangular algorithm,this paper introduces a series of steps such as tetrahedral verification,mirror detection and other star point verification,and makes the recognition rate reach 99.2%.Then,The algorithm of the algorithm is optimized by analyzing the detailed process of the algorithm,and the algorithm is optimized from the aspects of data storage,algorithm pruning and data structure replacement.The five versions are identified in detail.Algorithm,based on the triangular matching of the star map recognition algorithm has been optimized to a relatively high level,posture update rate has reached 33 Hz,in the same platform,with the star scale in the advanced ranks.Then,we try to apply the traditional star map algorithm to TI's DaVinci heterogeneous dual-core processor,make full use of DSP's real-time computing power,and improve the efficiency of the algorithm from the hardware level under the premise of maintaining low power consumption.From the environment to the bottom Principles have been carried out in-depth study of similar algorithms in the future transplant work has a very good guiding role.Finally,although the efficiency of the algorithm in DSP is not up to the expected efficiency,but the comparative analysis,such as the small amount of algorithm can not fully exploit the advantages of DSP performance,the algorithm jumps too much cause the DSP pipeline to clear the processor performance,fixed-point DSP Is not suitable for running the core data for the floating-point data algorithm and a series of useful conclusions for the future development of DSP algorithm or expect to use DSP to accelerate the algorithm has a good reference.