MATLAB implementation of remote sensing image clas
4.0
K-means algorithm is a typical point-by-point modified dynamically iterative clustering algorithms, is also a commonly used method, which is based on squared error criterion function. General practice is to select a number of representative points according to certain principles as the core of the cluster, and then put the rest of the points in a certain way (criterion) to go in and completes the initial classification. The initial classification is complete, recalculate the cluster Center , completed the first iteration. Then modify the cluster centers, so that for the next iteration. This change, there are two programmes, namely by modifying and lot-by-lot changes. Centre is a point-by-point modified class pixel samples belonging to a group according to a certain rule after the class, it is necessary to recalculate the mean value for this set of class