基于偏正态分布的星点细分定位方法研究  

Research on Star Subdivision Location Method Based on Skewed Normal Distribution

作　　者：贾瑞明[1] 马晓蕾[1] 郝云彩[2] 

机构地区：[1]北方工业大学电子信息工程学院,北京100144 [2]北京控制工程研究所空间智能控制技术国家重点实验室,北京100190

出　　处：《激光与光电子学进展》2016年第5期134-141,共8页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(61174004)

摘　　要：为了解决星点在实际成像过程中能量分布的偏正态分布问题,提高星点中心的定位精度,提出了基于偏正态分布模型的点扩展函数(PSF)相关算法。该算法根据实际星图中星点的能量分布特征,建立与之相对应的PSF,利用相关的原理,找到与星点能量分布相似度最高的PSF,通过确定对应PSF的最大值位置实现对星点中心的定位。实验结果表明,在星图噪声为N(0,0.001)的仿真条件下,且星点中心在1pixel内随机分布时,偏正态PSF相关法的星点中心平均定位精度可达到0.04pixel,远小于质心法0.4pixel和高斯曲面拟合法的1.03pixel。由实验结果可知,该算法定位精度高于质心法和高斯曲面拟合法,具有较好的抗噪声性和稳定性,提高了星点中心定位精度。In order to solve the energy distribution of the skewed normal distribution problems in the actual star imaging process, and improve the positioning accuracy of star point center, a point spread function （PSF） correlation algorithm based on skewed normal distribution is presented. The proposed algorithm establishes the corresponding PSF according to the energy distribution of actual star point. It uses the relevant principles to find the PSF with the highest similarity with star energy distribution, and then by determining the corresponding PSF maximum position to realize the positioning of the central star. The experimental results show that under the simulating condition of star image Gaussian noise in N（0,0.001） and the star center random distribution within 1 pixel, the average positioning accuracy of skewed normal distribution PSF correlation method can reach 0.04 pixel, which is far less than 0.4 pixel from the centroid method and 1.03 pixel from the Gaussian surface fitting method. Experimental results show that the proposed algorithm is better than the centroid method and Gaussian surface fitting method, which has good anti-noise performance and stability, and improves the positioning accuracy of the star center.

关 键 词：图像处理 偏正态分布 点扩展函数相关法 星点能量分布 质心法 高斯曲面拟合法 

分 类 号：V448[航空宇航科学与技术—飞行器设计]