为了提高两步相移干涉测量术的相位重建精度,提出了一种基于分区背景估计的相位盲解调算法。该算法选取干涉图中的一系列零值点作为种子,围绕种子生成Voronoi图,并将干涉图划分为若干区域,将区域内全部点背景设置为种子点背景,最后拼接为完整背景。对该算法进行了数值模拟和实验分析。数值模拟结果表明,理想情况下该算法解调相位的均方根误差(RMSE)接近实际背景法。实验结果表明,相比传统低通滤波方法,该算法解调相位的RMSE降低了16%。 In order to improve the phase reconstruction accuracy of two-step phase-shift interferometry, a phase blind demodulation algorithm based on partition background estimation is proposed. The algorithm selects a series of zero-value points in the interferogram as a seed, generates Voronoi diagram around the seed, divides the interference map into several regions, sets the background of all the points in the region as the seed point background, and finally splices into a complete background. The algorithm is simulated and analyzed experimentally. The numerical simulation results show that the RMSE of demodulation phase of this algorithm is close to the actual background method under ideal conditions. The experimental results show that the RMSE of the demodulated phase is reduced by 16% compared with the traditional low-pass filtering method.