基于联合双边滤波的高光谱遥感图像分类

Hyperspectral Remote Sensing Image Classification Based on Joint Bilateral Filtering

针对复合核在利用空间信息时使用固定大小的邻域会导致位于边缘处的像元模糊,从而导致边缘处容易错误分类的情况,提出了基于联合双边滤波器的高光谱图像分类方法。利用主成分分析提取出第一主成分作为滤波参考图像,接着对原始光谱图像联合双边滤波,使用支持向量机分类。实验结果表明本文提出的方法结果比复合核分类方法更好,并且比先分类后滤波的方式效果好。

For the composite kernel using fixed size neighborhood in utilizing spatial information, resulting in fuzzy at the edges and thus leading pixels at edges easy to be classified by mistake, we proposed a new method based on joint bilateral filter. Firstly, the first principal component is extracted by the principal component analysis as the filtered reference image, and then the original spectral image is filtered by joint bilateral filtering, finally a support vector machine is used to classify the filtered image. Experimental results show that the method proposed in this paper is better than that of the composite kernel classifica- tion method, and is better than the method of filtered after classification.