Shack-Hartmann (S-H) 波前传感器光斑图像阈值的选取对质心探测精度有较大的影响。本文分析了S-H波前传感器光斑图像的特点,研究了光斑图像阈值对质心探测精度的影响，在此基础上提出了一种基于灰度直方图原理的多峰自适应阈值选取方法。实验结果表明，对于信噪比＞4的S-H波前传感器光斑图像，本方法能够比较有效地自动选取每帧图像的阈值；但对于信噪比＜4的图像，分割效果不是很理想。将使用本阈值算法的S-H波前传感器应用到自适应光学系统中进行闭环校正实验，结果显示，校正后的系统波前误差从10.41λ(PV值，λ=632.8 nm)降低到0.12λ，基本达到了衍射极限水平，表明本方法可以满足S-H波前传感器的实用要求。

The threshold selection method for the spot image of Shack-Hartmann Wavefront Sensor (SH WFS) has a large influence on the centroid detection accuracy. This paper analyzes the features of spot image of a SH WFS,and studies the effect of the threshold of spot image on the centroid detection accuracy,and then it presents an adaptive threshold selection method based on gray histogram principles. The experimental results show that the method can effectively and automatically select the threshold of each frame image when signal-to-noise ratio(SNR) of the spot image of SH WFS is more than 4, but it can not get the ideal segmented image when the SNR is less than 4. The SH WFS with the threshold selection method is used in the adaptive optics system to carry out a closed-loop correction experiment. After correction, the wavefront distortion of 10.41λ is corrected to 0.12λ in PV (λ=632.8 nm), and the system resolution reaches the diffraction limit. These results show the method can meet the practical requirements of SH WFSs.