一种基于 Non-Local Means 的地基差分干涉雷达相位滤波改进方法

The phase pattern of ground-based differential interferometer radar usually contains a lot of phase noise，which seriously affects the phase unwrapping and deformation measurement results. In order to obtain a higher image quality，an improved Non-Local Means combined filtering algorithm is proposed in this paper. Firstly，the algorithm uses the coherence coefficient to construct an adaptive smoothing parameter model，which effectively improves the limitations of the Non-Local Means algorithm in the selection of filtering parameters. Secondly，since Wiener filtering can effectively filter the space-variant noise，this paper effectively combines adaptive Non-Local Means with Wiener filtering to better maintain the phase continuity and improve the image edge ambiguity. The measured data show that，compared with the adaptive median filtering algorithm，Goldstein algorithm and traditional Non-Local Means algorithm，the proposed algorithm has obvious advantages in suppressing phase noise，maintaining phase continuity and improving image edge ambiguity.