在光学镜面相位恢复检测中,由于CCD相机的横向失调和光路中倾斜误差的存在,采集图像会发生错位,影响检测结果的准确性。为了克服光路校准带来的困难,基于快速傅里叶变换图像配准算法以及梯度相位恢复算法,提出了亚像素图像配准相位恢复 (SIRPR) 算法。通过仿真,验证了该算法对于光路的横向偏移和镜面倾斜具有良好的矫正作用。对一块直径为110 mm的凹面反射镜进行实验测量,并利用SIRPR算法处理含明显错位光斑的光强图,将计算获得的面形结果与干涉仪直接测量的结果进行对比,面形之差的方均根值为0.1047λ(λ为波长),验证了算法的准确性。

In the phase recovery detection of optical surfaces, the collected image will be misaligned due to the lateral misalignment of the CCD camera and the tilt error in the optical path, which affects the accuracy of the detection results. In order to overcome the difficulty in optical path calibration, a subpixel image registration phase retrieval (SIRPR) algorithm is proposed based on the fast Fourier transform image registration algorithm and the gradient phase retrieval algorithm. The simulation results show that the proposed algorithm can correct the transverse deviation and mirror tilt of an optical path. A concave mirror with a diameter of 110 mm is measured experimentally and the SIRPR algorithm is used to process the light intensity graph with obvious dislocation spots. The calculated surface shape results are compared with those directly measured by the interferometer. The root-mean-square value of the surface shape difference is 0.1047λ(λ is wavelength), which verifies the accuracy of the proposed algorithm.