弱信号哈特曼波前探测算法的研究

发布时间：2018-07-18 15:47

【摘要】：自适应光学技术已经广泛应用于大口径地基望远镜系统中,用于克服大气湍流引起的波前畸变。波前校正效果直接受自适应系统的波前探测精度影响。本论文对液晶自适应光学系统中的弱信号哈特曼波前探测算法进行了深入研究。光斑质心探测精度是哈特曼波前探测器最重要的性能指标,它决定了哈特曼波前探测器的探测能力。在进行弱信号探测时光子噪声严重影响了质心探测精度。考虑到光子噪声在几何位置上重叠于信号,利用足够小的几何区域内光能量变化不大、而光子噪声则起伏较大的预想,提出了基于局部均值技术和加权技术的局部均值加权算法抑制光子噪声的影响。在子孔径光子数为100(5.5星等)、大气相干长度分别为10cm和5cm的观测情况下,采用局部均值加权算法抑制光子噪声后,使质心探测误差的RMS值分别由传统重心法的0.142?和0.181?减少到0.112?和0.145?,均减少了???。?对于大气湍流引起的波前畸变,Zernike模式系数并不是统计独立的,因此并不是最有效的重构模式。相比Zernike模式,利用系数统计独立的K-L模式进行波前重构能提高波前重构精度。在子孔径数20×20、大气相干长度10cm、质心探测误差0.1?的情况下,波前重构误差RMS值可以由??????减小到??????,减小了???。不同观测条件下最佳的重构模式数不同,模式数过多或者过少都会导致波前重构误差增大。因此提出采用差分星点像运动法从哈特曼探测的光斑质心阵列数据中同时统计出大气相干长度和质心探测误差,继而获得最佳重构模式数的方法。为了保证湍流的各态历经,同时考虑计算量,将该估计方法中子孔径中心间隔确定为1个子孔径,将采样时间长度和采样时间间隔分别确定为5000ms和50ms。即使在质心探测误差0.25?的情况下,根据该方法统计的大气相干长度和质心探测误差也能够使波前重构误差达到最小。液晶自适应光学系统采用哈特曼探测器进行倾斜像差探测,省却了专门的倾斜探测器,简化自适应光学系统的同时也提高了系统的能量利用率。将所有有效光斑质心偏移平均作为倾斜信号的小光斑法探测精度高,但动态范围小,而将所有光斑的重心偏移作为倾斜信号的大光斑法虽然动态范围可以扩大到整个CCD面板,但是探测误差又太大。为了能够对大振幅倾斜进行高精度校正,提出通过网格划分识别出子光斑和通过模板匹配将子光斑与其对应的微透镜进行关联的方法。该方法动态范围与大光斑法相同,而精度与小光斑法相同,即使在子孔径光子数100(5.5星等)、大气相干长度5cm的观测条件下,该方法依然能够保证倾斜回路闭环,闭环后倾斜探测误差PV值为0.07″。总之,本论文进行的弱信号哈特曼波前探测算法的研究不仅提升了自适应系统校正效果,而且使得自适应系统能够在更加极端的观测条件下工作。
[Abstract]:Adaptive optics has been widely used in large aperture ground-based telescope systems to overcome wavefront distortion caused by atmospheric turbulence. The effect of wavefront correction is directly affected by the detection accuracy of the adaptive system. In this paper, the weak signal Hartmann wavefront detection algorithm in liquid crystal adaptive optical system is studied. The accuracy of spot centroid detection is the most important performance index of Hartmann wavefront detector, which determines the detection ability of Hartmann wavefront detector. The accuracy of centroid detection is seriously affected by the noise at the time of weak signal detection. Considering that the photon noise overlaps the signal in a geometric position, the energy of light within a sufficiently small geometric region does not change much, while the photon noise is expected to fluctuate considerably, A local mean weighted algorithm based on local mean technique and weighted technique is proposed to suppress the influence of photon noise. When the photon number of sub-aperture is 100 (5.5 stars et al.) and the atmospheric coherent length is 10cm and 5cm respectively, the local mean weighted algorithm is used to suppress photon noise, and the RMSs of centroid detection error are changed from 0.142? And 0.181? Reduced to 0. 1 12? And 0.145? The Zernike model coefficients of wavefront distortion caused by atmospheric turbulence are not statistically independent and therefore are not the most effective reconstruction model. Compared with Zernike model, using K-L model, which is statistically independent, can improve the precision of wavefront reconstruction. At the subaperture number of 20 脳 20, the atmospheric coherence length is 10 cm, the centroid detection error is 0.1? The RMS value of the wavefront reconstruction error can be determined by? To reduce to a small, reduced. The optimal number of reconstructed modes is different under different observation conditions. Too many or too few modes will lead to the increase of wavefront reconstruction error. Therefore, the differential star image motion method is proposed to calculate the atmospheric coherent length and centroid detection error simultaneously from Hartmann's centroid array data, and then obtain the best reconstruction mode number. In order to ensure the ergodic states of turbulence and take into account the computational complexity, the central interval of neutron aperture is determined to be one sub-aperture, and the sampling time length and sampling time interval are determined to be 5000ms and 50ms, respectively. Even if the centroid detection error is 0. 25? The atmospheric coherence length and centroid detection error calculated by this method can also minimize the wavefront reconstruction error. In liquid crystal adaptive optical system, Hartmann detector is used to detect skew aberration, which saves the special tilting detector, simplifies the adaptive optical system and improves the energy efficiency of the system. The small spot method, which uses the average centroid migration of all effective light spots as tilting signals, has high detection accuracy but small dynamic range. However, the large spot method, which uses the center of gravity of all light spots as tilting signals, can be extended to the whole CCD panel, although the dynamic range can be extended to the whole CCD panel. But the detection error is too big. In order to correct the large amplitude tilt accurately, a method is proposed to identify the sub-spot by mesh division and to associate the sub-spot with the corresponding microlens by template matching. The dynamic range of the method is the same as that of the large spot method, and the precision is the same as that of the small spot method. Even under the condition of the photon number of sub-aperture 100 (5.5 stars, etc.) and the atmospheric coherent length 5cm, the method can still guarantee the closed loop of the tilt loop. The detection error PV is 0.07 ". In a word, the research of weak signal Hartmann wavefront detection algorithm in this paper not only improves the correction effect of adaptive system, but also enables the adaptive system to work under more extreme observation conditions.
【学位授予单位】：中国科学院研究生院(长春光学精密机械与物理研究所)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：O439

【相似文献】