星敏感器星图降噪预处理及其硬件实现

发布时间：2019-01-05 04:34

【摘要】：目前利用星敏感器对航天器进行姿态测量是精度最高的。相比较于其它类型的姿态敏感器,由于星敏感器的自主性强、无姿态积累误差等特点,使得飞行器的姿态确定可以在无陀螺的条件下进行。星点坐标提取是星敏感器研制过程中的关键步骤之一。恒星星像坐标的提取精度是衡量星敏感器性能的重要指标。本学位论文将从星图降噪预处理和提高恒星星像坐标提取精度等问题出发,对星图滤波算法进行深入研究。以分析国内外研究发展现状为基础,完成了以下工作:本文阐述了星敏感器的工作原理和硬件结构,同时对星敏感器的姿态测量精度进行了分析。列举了星敏感器的主要精度指标。分析了影响星敏感器姿态测量精度的因素,包括传感器像元尺寸的限制和测量中引起的误差,对于后者,主要分为三类,即光学系统误差、算法误差、噪声引起的误差。然后阐明了针对上述引入的误差提高星敏感器精度的方法,包括先进的标定技术、优化算法、以及抑制噪声和对星图进行降噪预处理等。本文对比了几种滤波方法,选取了更适合对星图滤波的二维FIR滤波作为星图的预处理方法。分别用窗函数法和频率取样法设计了25(5?5)阶的二维FIR滤波器,并将得到的滤波器参数分别用上位机对星图进行滤波处理,对滤波前后的星图进行频谱分析,分析滤波效果。并对这两种方法设计的滤波器滤波后的星图进行频谱分析,对比滤波效果。为了更好的说明滤波器阶数对滤波效果的影响,本文用频率取样法设计了9(3?3)阶的二维FIR滤波器,并与频率取样法设计的25(5?5)阶的二维FIR滤波器的滤波效果进行对比分析,进而得出结论。根据理论分析和仿真实验,本课题选用频率取样法设计的25阶(5?5)二维FIR滤波器的参数,应用VHDL语言进行滤波模块的编写,将所编写的模块调试成功后整合到系统中,进行星图的降噪预处理。将FPGA提取的星像坐标上传至上位机,将坐标提取结果与未加入FIR滤波模块的坐标提取结果进行对比,对所得到的实验数据进行分析。实验结果表明对星图滤波之后对高频噪声滤除效果显著,恒星星像坐标的提取精度有所提升。
[Abstract]:At present, the attitude measurement of spacecraft using star sensor is the most accurate. Compared with other types of attitude sensors the attitude determination of the spacecraft can be carried out without gyroscope due to its strong autonomy and no attitude accumulation error. Star coordinate extraction is one of the key steps in the development of star sensor. The precision of constant star image coordinate extraction is an important index to evaluate the performance of star sensor. In this dissertation, the star image filtering algorithm is studied based on the preprocessing of star image denoising and improving the precision of star image coordinate extraction. Based on the analysis of the present research and development situation at home and abroad, the following work has been accomplished: in this paper, the working principle and hardware structure of the star sensor are described, and the attitude measurement accuracy of the satellite sensor is analyzed. The main precision indexes of star sensor are listed. The factors that affect the precision of attitude measurement of star sensor are analyzed, including the limitation of sensor pixel size and the error caused by measurement. For the latter, there are mainly three kinds of errors, namely, optical system error, algorithm error and noise error. Then, the methods to improve the accuracy of star sensor are expounded, including advanced calibration technology, optimization algorithm, noise suppression and pre-processing of star map. In this paper, several filtering methods are compared, and two dimensional FIR filter, which is more suitable for star map filtering, is selected as the preprocessing method of star map. Two dimensional FIR filters of order 25 (5 ~ 5) are designed by using window function method and frequency sampling method respectively. The parameters of the filter are processed by the upper computer, and the spectrum of the star map before and after filtering is analyzed. The filtering effect is analyzed. The filtered star map of the filter designed by these two methods is analyzed by spectrum analysis, and the filtering effect is compared. In order to better explain the influence of filter order on filtering effect, a 2-D FIR filter of order 9 (3 ~ 3) is designed by frequency sampling method. The filter effect of 25 (5 ~ 5) order 2-D FIR filter designed by frequency sampling method is compared and analyzed, and a conclusion is drawn. According to the theoretical analysis and simulation experiment, the parameters of 25 order (5 ~ 5) 2-D FIR filter designed by frequency sampling method are selected in this paper. The filter module is compiled by VHDL language, and the module is successfully debugged and integrated into the system. The star map denoising is preprocessed. The satellite image coordinate extracted by FPGA is uploaded to the upper computer, and the result of coordinate extraction is compared with that of the coordinate extraction without FIR filter module, and the experimental data obtained are analyzed. The experimental results show that after the star map filtering, the high frequency noise filtering effect is remarkable, and the extraction accuracy of the constant star image coordinates is improved.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V448.22;TP391.41

【参考文献】