相位解缠面临的困难之一是相位的不一致性。噪声和采样率是导致相位不一致的两个根源。目前, 对于相位不一致点的处理是, 不管其产生的根源如何, 均采取绕过这些点的办法。本文提出一种新的相位解缠方法。它首先设法利用由于采样率所引起的相位的不一致性, 进行相位梯度场估值, 然后用非加权最小二乘法求一次相位, 以此相位为初值, 最后用加权最小二乘法, 迭代求相位。由于它赋予加权最小二乘法一个比较合理的初值, 使得迭代速度和精度大为提高, 模拟结果证实了这一点。

An inconsistency is one of the engineering obstacles to unwrap phase. Both noise and undersampling can cause inconsistencies. Up to now, the method of processing inconsistencies is to avoid inconsistent pixels whether they are caused by noise or undersampling. This paper presents a new method for phase unwrapping. After phase-gradient is estimated, unweighted and weighted phase unwrapping are used. It has been shown that the precision of phase can be improved because of a reasonable initial quantity for weighted least-squares.