光学联合变换相关器因其高速并行光学运算特性在实时图像位移精密测量系统中有广泛的应用前景，然而位移测量亚像素误差是制约其精度的瓶颈。据此提出了一种在图像位移精密测量应用中提取相关峰位置的最优算法，使得系统亚像素误差达到最小，并研究了亚像素误差的补偿方法。通过仿真和实验分析，对比加权质心提取算法、抛物面拟合算法和高斯面拟合算法对亚像素误差的影响，得到最佳的相关峰位置提取算法是权值次数为2的质心提取算法；提出的亚像素定位补偿算法可将亚像素误差由±0.1 pixel减小到±0.04 pixel，大大降低了亚像素误差对位移测量系统的影响，提高了光学联合变换相关器位移的测量精度。

The accuracy displacement measurement system based on an optical joint transform correlator has widely potential application for its high speed and parallel computing.But its accuracy is restricted because of the centroid subpixel error. An optimal correlation peak extraction algorithm and a compensation algorithm, which can be applied to the practical displacement measurement with the minor systematic error, are presented. Though simulation, experimental analysis and comparing the subpixel error effects based on weighted center of mass (COM), paraboloid fitting algorithms, Gaussian fitting, the best correlation peak extraction algorithm is COM with a weight degree of 2. The experimental data indicates that the residual error is declined from ±0.1 pixel to ±0.04 pixel through the new compensation algorithm, and the effection of subpixel error of the system is declined accordingly, and the displacement measurement accuracy of an optical joint transform correlator is improved.