【摘要】 针对空间遥感ＴＭ图象和ＳＡＲ图象信息的特征层融合应用于地物分类，提出了一种结合Ｍａｒｋｏｖ随机场和ＢＰ神经网络模型的多源遥感图象迭代分类方法。该分类方法与现有的基于Ｍａｒｋｏｖ模型的分类器相比具有无须假设条件概率密度函数模型的优点；与ＢＰ神经网络分类器相比，由于其考虑了类别标号的空间相关性，提高了分类精度；有别于传统的上下文分类器：它是通过迭代过程来实现分类的，在考虑了类别标号的空间相关性的同时考虑了象素类别的特征属性更多还原

【Abstract】 An iterative technique of multisource remote sensing image classif ication based on Markov random fields and BP neural networks is presented in accordance with the character fusion classification of space remote sensing TM image and SAR ima ge. Compared with the available classified method based on Markov random fields, its merit shows that it is not necessary to assume the conditional probability function, and it has higher classified accuracy than the methods based on BP neu ral networks for considering the space correlation of class lable. The differenc e between this classificator and the traditional context one is that the former implements classification through an iterative progress in which the character a ttribute of pixel category as well as the space correlation of class lable is co nsidered.更多还原