基于模糊高斯基函数神经网络的遥感图像分类

针对遥感图像分类的特点，提出了一种基于模糊高斯基函数神经网络的遥感图像分类器。该分类器将模糊技术与神经网络相结合，采用神经网络来实现模糊推理，利用神经网络的学习能力来达到调整模糊隶属函数和模型规则的目的，从而使系统具备了自适应的特性，实验结果表明，这种基于模糊高斯基孙数神经网络的分类器经过训练后，可应用于遥感图像的分类，其分类精度明显高于传统的最大似然分类法。

Remote Sensing Image Classification Based On Fuzzy Gauss Function Neural Network

Information on land\|cover/land\|use gives important clues for both more effective planning and the wiser exploitation of natural resources. Get useful information from remote sensing image of large areas is a time\|consuming process. Computer\|aided classification has provided an alternate, effective method. Improving classification accuracy is a key issue in remote sensing image classification. The main drawback of traditional remote sensing information classification methods is its low precision. In recent years the artificial neural network technology and fuzzy technology have been developed and applied to remote sensing data classification problem. Compared to the conventional statistical classifier, the neural network classifier is a non\|parametric method and has the capacity of self\|learning and high robust. The fuzzy technology has the superiority in logical inference and can be used in high level information processing. In this paper, considering the features of remote sensing images, we synthesize the advantages of these two methods and proposed a remote sensing image classifier using fuzzy gauss function neural network. In the system, fuzzy technique and neural network technique are combined, and the fuzzy inference is realized by neural network. It using the learning ability of neural network to adjust the membership function and fuzzy rule, which endue the system with the distinguished capability of self\|adapt and self\|learning. We applied the proposed method to construct a classifier for remote sensing image classification experiment research. Experimental results show that the fuzzy gauss function neural network classifier can be used in remote sensing image classification, and its classification precision is superior to that of the conventional maximum\|likelihood.