关键词: SAR影像, 去噪, 直方图, 偏微分方程, 图像增强, ROF

Abstract:

In this paper, an improved algorithm based on histogram partial differential equalization(PDE) is proposed. This algorithm overcomes the problem of information damage in detail region, caused by eliminating speckle noise in existing SAR image enhancement algorithm. The method is based on the original histogram equalization PDE model, and combined with the ROF (Rudin Osher Fatemi) denoising model, the diffusion coefficient is controlled by the image gradient of edge features, and is kept small in edge region to retain edge information while big in non-edge region to remove noise. Then self-adaptive correction is applied to the denoised image, to suppress the unevenness in smoothing area of the image caused by ladder effect. The advantage of the algorithm is that de-noising and image enhancement are carried out simultaneously, so as to reduce the influence of coherent noise while preserving the contours of the edge of the feature. In order to verify the applicability of the algorithm, Sentinel-1A data is selected for example test. The results show that this method can effectively remove the coherent noise in the smooth region, and the information in edge and detail region is well preserved.

Key words: SAR images, de-noising, histogram, PDE, image enhancement, ROF