为提高哈特曼夏克波前传感器（HS-WFS）的光斑质心探测精度以实现光学系统的高精度波前检测，提出了一种有效的质心探测方法。该方法利用非线性滤波和窗口法对整幅光斑图像进行全局处理后，结合中值滤波、三次样条插值和自适应Otsu阈值法对单个光斑进行局部处理。分析了三次样条灰度插值点个数不同，探测精度和计算时间的变化规律。采用该方法探测了含有噪声的光斑图像，其质心探测误差仅为0.0442 pixel，比传统的非线性滤波、Otsu阈值法和探测窗口法探测精度分别提高了91.86%、87.97%和31.79%。对已知波像差的光学系统进行了仿真检测，得到的波前检测精度峰谷(P-V)值为0.0098 λ，精度均方根(RMS)值达到0.0027 λ。结果表明该方法能够提高质心探测精度，可用于高精度光学系统的检测。

In order to improve spots centroid detection accuracy of Hartmann-Shack wavefront sensor (HS-WFS) to achieve high-precision of wavefront test of optical system, a centroid detecting method is proposed. The nonlinear filtering and the windowing spot image in global processing are used, combining with median filtering, cubic spline interpolation and adaptive Otsu thresholding for a single spot in local processing. The regulation of centriod error and calculation time with different interpolation points is discussed. For a spot image with noise, the centroid detection error is only 0.0442 pixel. Compared with the traditional methods of nonlinear filtering, the Otsu thresholding method and the detection windowing, the accuracies are increased by 91.86%, 87.97% and 31.79%. The known wavefront aberration of an optical system is simulation tested. The wavefront testing accuracy is 0.0098 λ in peak-valley (PV) value, and is 0.0027 λ in root-mean-square (RMS) value. The results show that this method enhances the accuracy of centroid detection, and is suitable for high-precision optical systems.