基于混合核函数的极限学习机遥感图像分类

Remote Sensing Image Classification Based on Hybrid Kernels Function for Extreme Learning Machine

针对遥感影像数据集庞大,地物复杂难辨等特性导致分类难度加大的问题,文中构建了一种基于混合核函数极限学习机的遥感图像分类方法。运用该方法对遥感图像数据集进行分类处理,并将其与单核极限学习机、无核极限学习机、支持向量机等方法进行了对比。实验结果表明,基于混合核函数的极限学习机在对遥感图像进行分类时,其总体精度更优,且一致性效果更好。

Characters of large data sets,surface features complex etc made it difficult to classify remote sensing image. This paper constructed a extreme learning machine classifier with hybrid kernels function and then put up a remote sensing image classification method based on the extreme learning machine classifier with hybrid kernels function. This paper used the method to classified the data sets of the remote sensing images.By comparing the extreme learning machine classifier with single kernel function learning machine,extreme learning machine and the support vector machine about coping with remote sensing image data sets,the experimental results indicate that the extreme learning machine using hybrid kernels function has higher overall accuracy and better consistency.