An Efficient Phase Unwrapping Method Based on Unscented Kalman Filter

Abstract

In this article, we proposed a phase unwrapping (PU) method which combines with unscented Kalman filter, pixel classification, and an efficient path-following strategy. The characteristics of the proposed method are summarized as: 1) the path-following strategy speeds up the process of PU without decreasing the accuracy; 2) the reliability of each pixel will be graded according to the position of residue and pixel classification strategy; and 3) different from the traditional methods, the proposed method can perform filtering and PU at the same time to prevent global propagation of error. In addition, we also introduce a signal model which can obtain a similar correlation map by only using a wrapped phase image when without the primary-secondary image. The results on synthetic data and real data show that the proposed method can obtain better results.