研究了亚像素数字散斑相关测量中整像素搜索窗口、拟合窗口、散斑尺寸等因素对曲面拟合法测量精度的影响。采用二元二次多项式曲面拟合方法，利用计算机生成的模拟散斑图像，对水平位移分别为0.1，0.3，0.5，0.7，0.9 pixel的图像进行了研究。结果表明，最佳搜索窗口为31×31～51×51 pixels，最佳拟合窗口为3×3 pixels，散斑图的平均散斑尺寸为2 pixel时，能较容易地实现相对误差小于5%的目标。而在高斯噪声免疫力方面，0.9 pixel以上的位移具有较强的免疫力。在散斑相关测量精度要求不是很高(>1 μm)的情况下，曲面拟合法是一种非常适合的方法。

The influence of registration subset size, fitting window size， and speckle size on the accuracy of the curved surface fitting method are studied in the sub-pixel displacement digital speckle correlation measurement. The speckle patterns generated by computer with displacement of 0.1, 0.3, 0.5, 0.7, and 0.9 pixel are calculated using binary quadratic polynomial curved surface fitting method. It shows that the optimal parameters of registration subset size, fitting window size, and speckle size are 31×31～51×51 pixels, 3×3 pixels, and 2 pixel, respectively, and it is easy to measure the target with relative error less than 5%. When the displacement is above 0.9 pixel, the calculation method is immune to Gaussian noise. So the curved surface fitting method is effective when the requirement of accuracy is not critical (>1 μm).