【摘要】 研究了一种基于模糊量化与多层前馈神经网络相结合进行多类别遥感图像植被分类的新方法.对分类指标 进行π隶属函数模糊量化之后,进行多层前馈神经网络(MLP)训练,再用训练后的MLP网络构成监督分类器,并 以宁波市彩红外航空遥感图像资料为数据源,选取园林区为试验区,对植被进行监督分类,分类精度达78%.研究 成果表明,该方法适用于多类别的遥感图像植被分类,其分类效果也较好.更多还原

【Abstract】 A new method, which is constructed by Fuzzy Quantification and Multi-layer Feed-forward Neural Networks, is put forward for multi-category remote sensing image classification of vegetation. First, indexes of vegetation classification are fuzzily quantified by π function. Secondly, training samples are used to train the multi-layer feed-forward neural networks. At last, a supervised classifier is constructed and applied for vegetation classification from aerial image. Arial infrared photo of Ningbo city is choose as testing image and classification precision is 78%. The experimental results show that the approach is suitable for vegetation classification from multi-category remote sensing images.更多还原