中文说明:根据实验数据设计K均值聚类分析分类器,实验数据采用遥感彩色图像,以图像的所有象素为样本集,每一象素点的R、G、B值作为其特征向量。1)选择合适的类别数K和初始聚类中心。2)选择距离测度。3)设计迭代中止条件,或人为设定迭代次数。
English Description:
K-means cluster analysis based on experimental data design classifier, experimental data of remote sensing color images, all pixels of the image for the dataset, the r, g, b values for each pixel as its eigenvectors. 1) select suitable category number k, and the initial cluster centers. 2) select a distance measure. 3) termination of design iteration, or artificially set number of iterations. 4) loop iteration ends, all pixels of the r, g, b values in various centers of r, g, b values, draw a classification diagram. 5) analysis of different initial cluster centers and the iteration condition influence classification results.