多尺度多特征融合的高分辨率遥感影像分类

针对高分辨率遥感影像多尺度、空间分布复杂以及特征繁多的特点，从遥感影像特征提取的尺度效应以及各类地物显著性特征各异入手，提出了基于多尺度多特征融合的高分辨率遥感影像分类的方法。该方法构建最优尺度分割函数模型，寻找出各地物的最优尺度，分别提取影像的纹理、颜色和形状特征。在此基础上，利用各地物特征的显著性差异实现多尺度下多特征的加权融合。该加权融合方法突破了常规的最优尺度分割算法未能充分考虑各类地物特征差异性的局限性，通过分析各类地物的显著性，建立了各个特征在分类中所占权重的模型。实验结果表明：相对传统无监督分类算法，该方法准确率提高约7%，且运行效率高。

High resolution remote sensing image classification based on multi-scale and multi-feature fusion

In view of the high resolution remote sensing image with multi-scale,complex spatial distribution and the characteristics of a wide range of features,the method of high resolution remote sensing image classification is proposed based on multi-scale and multi-feature fusion,which is starting with the scale effect of feature extraction from remote sensing image and various conspicuousness of different objects.The optimal segmentation scale function is constructed using the method.The optimal scales of different objects are obtained,and texture,color and shape features are extracted respectively.The multi-scale and multi-feature weighted fusion is realized by using significant differences of different objects in characteristics.The weighted fusion method breaks through the limitation of the conventional optimal scale segmentation algorithm,which fails to fully consider the diversity of all kinds of features of different objects.By analyzing the significance of all kinds of features,a model is established based on the weight of each feature.Experimental results show that the accuracy of this method is increased by about 7％ compared with that of the traditional unsupervised classification algorithms,and the operation efficiency is high.