FasART模糊神经网络用于遥感图象监督分类的研究

说明了遥感图象数据的非线性性质，目视的图象分类实践是一个模糊推理的过程，模糊神经网络遥感图象分类符合其事物的内在规律，具有理论优势，分析了模糊ART，模糊ARTMAP和FasART模型的结构和原理，详细地阐述了FasART是一种基于模糊逻辑系统的神经网络，提出了一种简化的FasART模型，改变了一般遥感数据的模糊化方法，采用中巴资源一号卫星数据进行测试实验，结果表明，该简化的FasART模型能用于遥感图象的监督分类，其分类精度高于模糊ARTMAP神经网络和K均值算法，且性能稳定，有较好的抗干扰能力，尤其具有良好的处理两组相似程度比较接近的，和同组数据模式变化较大的非线性数据的能力。

A Study of FasART Neuro-fuzzy Networks for Supervised Classification of Remotely Sensed Images

The paper explains briefly that the remotely sensed data is non linear, and the practice of its classification by mans eyes is a process of the fuzzy inference. The fuzzy neural networks has a theory dominance, because it accords with the nature rule of classification of remotely sensed images. Analyses the architecture and principles of fuzzy ART, fuzzy ARTMAP. Discusses in detail that FasART is a neural networks based on fuzzy logic system. Put forward a simplified FasART architecture and change the general method of remotely sensed data fuzzification. With the testing of the CBERS -1 data, the results declares that the simple FasART model can be used to supervised classification of the remotely sensed images. The precision of the classification is higher than that of fuzzy ARTMAP and K means. The classification of FasART model has better stabilization and anti jamming, and has capability of dealing with non linear data especially.