光电器件的CRLB特性及定位技术的误差比较  

CRLB Feature of Optoelectronic Device and a Comparison of the Locating Errors for Centroiding Algorithms

作　　者：张俊[1] 郝云彩[1] 龙也[1] 刘达[1] 

机构地区：[1]北京控制工程研究所,北京100080

出　　处：《光学学报》2015年第8期310-317,共8页Acta Optica Sinica

基　　金：自然科学基金(61174004;61302177)

摘　　要：含有噪声的光电成像器件定位过程中,Cramer-Rao Lower Bound(CRLB)被用来计算器件的定位精度,作为任何无偏估计的下限,CRLB可作为判据评估定位技术是否满足最小方差无偏(MVU)特性。本文给出了成像器件在泊松和高斯噪声下的CRLB特性,并据此研究质心法(COG)、迭代加权质心法(IWCOG)、最小二乘高斯拟合法(GLSF)的定位误差。理论分析和仿真验证均表明,IWCOG和GLSF定位误差与光电成像器件的CRLB相同,满足MVU特性,而COG定位误差最大;COG耗时最少,GLSF最耗时不能在轨应用而IWCOG用时仅比COG高几倍。充分证明迭代加权质心法具有低误差、实时和鲁棒特性,因而适用于星敏感器、Shack-Hartmann传感器波前定位等高精度定位场合。In the positioning procedure for point-like sources generated in optoelectronic devices, Cramer- Rao Lower Bound （CRLB） is used to estimate the locating accuracy of the devices contaminated with noise. As the lower bound of any unbiased estimator, CRLB is known as a criterion to assess whether an estimator meets the characteristics of minimum variance unbiased （MVU）. The CRLB characteristies on the performance of the imaging device under both Poisson and Gaussian noise are given and accordingly the positioning error of three current centroiding algorithms, center of gravity （COG）, iteratively weighted center of gravity （IWCOG） and Gaussian least squares fitting（GLSF）, is investigated respectively. Theoretical analysis and simulation results show that IWCOG and GLSF share the same position error and reach CRLB of optoelectronic imaging device, thus meet MVU features. That is to say, COG makes the biggest positioning error, followed by IWCOG and GLSF. And the computing time of IWCOG is just several times longer than COG algorithm, while GLSF algorithm is certainly the time-consuming procedure to make it inapplicable on-board on star tracker. IWCOG is proven to be as a method with low error, real-time and robust algorithm, making it possible for high- precision condition such as star tracker, Shack-Hartmann sensor.

关 键 词：光电子学光器件 质心法 迭代加权质心法 最小二乘高斯拟合法 定位误差 最小方差无偏 星敏感器 

分 类 号：V448.22[航空宇航科学与技术—飞行器设计]