一种采用自适应LPA-ICI的最小二乘法相位展开

PHASE UNWRAPPING USING ADAPTIVE LPA-ICI BASED ON LEAST SQUARE ALGORITHM

针对基于离散余弦变换的最小二乘法相位展开过度平滑噪声的问题,提出一种采用LPA-ICI算法去除折叠相位中噪声的解决方法。该方法获取折叠相位,根据LPA算法零阶近似和ICI准则来逐点自动调整窗口尺度,从而获取滤波窗口函数。一阶近似则利用该窗口函数进行相位去噪,再采用公认的最具鲁棒性的最小二乘法进行相位展开,获取整个相位场的绝对相位值。通过仿真和实验,与最小二乘法和PEARLS相位展开法对比,结果表明,该方法可以有效地对信噪比较低的折叠相位进行相位展开,具有良好的实时性和准确性。

Aiming at solving the problem of over-smoothing noise caused by the least squares phase unwrapping algorithm based on DCT, the LPA-ICI algorithm was proposed to be used to denoise. After obtaining the wrapped phase, the zero-order approximation, based on LPA, was used to select the adaptive window size pointwise with ICI rule. The first-order approximation was used to filter the phase in the window, until getting the satisfactory phase map. For the absolute phase, least squares phase unwrapping algorithm, well known as the most robust method, was applied to the denoised wrapped phase. Through simulation and experiment, our method was compared with the least squares method and PEARLS phase unwrapping method. The results show that this method can effectively unfold the phase of the folded phase with low SNR, and has good real-time and accuracy.