A New K-mean Color Image Segmentation With Cosine Distance for Satellite Images
This paper represents unsupervised method of k-means segmentation which is new adaptive technique of color-texture segmentation. With the progress in satellite images, the image segmentation technique for generating and updating geographical information are become more and more important. This algorithm first enhance the image then applying clustering based k-means segmentation technique, using Lab color space and using cosine distance matrices instead of sqeuclidean distance. With this it is possible to reduce computational time and calculation for every pixel in the image.