基于高斯低通滤波的超光谱遥感图像分类研究

在超光谱遥感图像的分类中,图像的类别可分性代表了图像的自然属性并决定了分类器能够达到的最优性能。在研究影响分类效果诸因素的基础上,提出了利用高斯低通滤波提高类别可分性的方法,在假设数据为多元正态分布的基础上,用Bhattacharyya距离衡量滤波前后样本集的类别可分性。在此基础上,构造了分类器,并进行了实际的分类测试。实验结果说明高斯低通滤波器能够提高类别可分性,因而能够提高分类精度。

Hyperspectral remote sensing classification based on Gaussian lowpass filter

In hyperspectral remote sensing image classification,class separability represents the nature of data set and decides the optimal performance of classifier.On the basis of researching factors effecting classification accuracy of hyperspectral remote sensing image,gaussian lowpass filter are used to improve class separability,Bhattacharyya distance under multi-dimension normal distribution are used to scale class separability before and after filtering.After these steps,classifier was constructed,experiments proved that gaussian low pass filter can increase class separability thereby increase classification accuracy.