A Speckle Reduction Method Based on Hyperspectral and SAR Image Fusion

SAR images are affected by speckle noise which needs to be suppressed. An ideal speckle reduction method must smooth noise and preserve detail both. However, there is a tradeoff between such two goals for many methods. This paper proposes a speckle reduction method based on hyperspectral and SAR image fusion. It combines the coherent portions from MNF transformation for hyperspectral image and the SAR image. After radiance correction on the united data, the Correlation Simulating Analysis Model (CSAM) judge based SRSSHF filter is used to smooth the image, and the dark isolated points regarded as the remaining noise are eliminated. Experimental results on Radarsat-1 imagery show that the proposed method in this paper smoothes the noise and keeps the features both well.