An Image Fusion Method of SAR and Optical Images, Based on Image Intensity Fields, by Reducing the Effect of Speckle Noise

Abstract

This study proposes an improved fusion method, that takes advantage of the combined strengths of existing fusion methods. First, current methods are compared using a fusion of noisy images from the Synthetic Aperture Radar (SAR) database, with optical images acquired at the same location and time. The obtained image and metric results showed that combining optical images with de-noised SAR provides better performance. Experiments have also shown that removing noise in SAR data causes the loss of important data in images. The proposed method divides the image into small patches in the noise removal phase. By calculating the standard deviation of these sub-patches, a different noise reduction ratio is applied for each region, thus preventing the loss of important detail features in the image. The proposed method has been compared with fusion methods recognized in the existing literature. Experimental results demonstrate that the proposed method, performs better than current fusion methods. The proposed method also yields better metric results, over other methods and it also eliminates the noise problems, often present in the images.