Channel Balance Algorithm Based on Two-Dimensional Gaussian Kernel Function and Non-Local Means Filter

Abstract

Multi-channel moving target detection algorithm is an essential technique for ground moving target indication (GMTI). The inconsistency of magnitude and phase characteristics between channels will affect the clutter suppression effect of the Synthetic Aperture Radar (SAR) system, and thus affect the final performance of moving target detection. To solve this problem, a channel balance algorithm based on the two-dimensional Gaussian kernel function [1] and the non-local means (NLM) filter is proposed in this paper. Firstly, the phase error is estimated and compensated by the two-dimensional Gaussian kernel function in the image domain, and then the magnitude error is compensated by the mean filtering. Finally, the NLM filter is applied to further suppress clutter. The proposed algorithm can effectively eliminate magnitude and phase errors, and improve the coherence between different channels. Experimental results on real airborne SAR data demonstrate the effectiveness of our algorithm.