The conventional approach of terrain image classification whichassigns a specific class for each pixel is inadequate because the areacovered by each pixel may embrace more than a single class. Fuzzy settheory which has been developed to deal with imprecise information canprovide a more appropriate solution to this problem. In the paper, theauthors used the fuzzy c-means clustering algorithm for the segmentationof a polarimetric SAR image. The distance measure utilized in thealgorithm was derived from the complex Wishart distribution of the pixeldata presented in the covariance matrix format. The algorithm computesthe feature covariance matrix for each class and generates a fuzzypartition of the whole image. Classification of the image is achievedusing a defuzzification criterion. The results are similar to those ofsupervised statistical methods. NASA/JPL AIRSAR data is used tosubstantiate this fuzzy classification algorithm
展开▼
