论文部分内容阅读
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
Aiming at the feature classification of spatial remote sensing TM image and SAR image information, a multi-source remote sensing image iterative classification method based on Markov random field and BP neural network model is proposed. Compared with the existing classifier based on Markov model, this classification method has the advantage of no need to assume the conditional probability density function model. Compared with the BP neural network classifier, it improves the classification accuracy by considering the spatial correlation of class labels ; Unlike the traditional context classifier: it is through the iterative process to achieve the classification, taking into account the spatial correlation of class labels at the same time taking into account the characteristics of the pixel category attributes