针对小波变换轮廓术中相位无法快速准确展开的问题, 提出了一种能有效抑制噪音的相位展开算法.首先, 分析了小波脊系数幅值质量图和相位导数方差质量图的特征;然后依据相位值与其数学期望的偏差程度, 利用小波脊系数幅值矩阵和导数方差思想构造了 “小波脊系数幅值导数方差”质量图;最后分别进行软件仿真和真实模型实验, 并将利用新质量图引导的相位展开效果与传统相位展开质量图法和改进的质量图法进行对比.实验结果表明, 本文算法提高了相位展开精确度, 相位展开误差率降低了2.61%.所提质量图易于构造, 在引导相位展开时不仅能得到较准确的绝对相位值, 还能有效抑制噪音的影响,从而重建出精确度高的物体三维模型.

In order to solve the problem that phase can’t be quickly and accurately unwarped in wavelet transform profilometry, a algorithm was presented which could effectively suppress noise. Firstly, the characteristics between the amplitude of wavelet ridge coefficient quality map and phase derivative variance quality map were analysized. Then, based on the deviation degree between phase vaule and its mathematical expectation, a “the amplitude of wavelet ridge coefficient variance derivative” quality map was constructed with the wavelet ridge amplitude coefficient variance matrix and variance derivative. Finally, from computer simulation and real experimental modeling, unwrapping phase based on the new quality map was compared with the traditional phase unwrapping method and the improved quality map method. Experimental results show that the accuracy of phase unwrapping is improved, the phase unwrapping error is decreased by 2.61 percent. The proposed quality map is construted quickly, phase unwrapping guided by the new quality map can obtain accurate absolute phase and effectively suppress noise, which can reconstruct accurate 3D model through the obtained absolute phase.