ISODATA的基本思想：算法运行前设定参数，在每轮迭代过程中，样本重新调整类别之后计算类内及类间有关参数，并和设定的门限比较，确定是两类合并为一类还是一类分裂为两类，不断地“自组织”，以达到在各参数满足设计要求条件下，使各模式到其类心的距离平方和最小。-The basic idea of ISODATA: Before running the algorithm to set the parameters in each iteration process, the sample readjust after intra-class computing category and type of parameters, and to compare and door limit set by determining the categories into one category or a class split into two groups, constantly　self-organization , in order to achieve the parameters meet the design requirements under the conditions of the respective mode to the square of the distance and the minimum class of their heart.