Single Look SAR Image Segmentation Using Local Entropy, Median Low-Pass Filter and Fuzzy Inference System

摘要: Synthetic Aperture Radar (SAR) image segmentation based on intensity information alone poses severe inaccuracy due to the presence of speckle noise. The multiplicative noise produces salt and pepper effect in end product thereby making it difficult segment images efficiently. Moreover, same reason, by applying classical methods which are thresholding clustering, satisfactory results often not achieved. Since SAR as they their original state cause a obstacle segmentation, here is proposed first extract roughness feature terms local entropy median filtered image; then feed fuzzy inference system. system takes inputs from two different features decide criteria. result Fuzzy classifier thresholded using Otsu’s method get final segmented image. simplifies generalizing possible spatial be extracted incorporated process while resulting efficient effective results.