基于SVM的高光谱遥感图像亚像元定位

发布时间：2018-01-13 08:21

本文关键词：基于SVM的高光谱遥感图像亚像元定位　出处：《武汉大学学报(信息科学版)》2017年02期 　论文类型：期刊论文

【摘要】：提出了基于支持向量机(support vector machine,SVM)的高光谱遥感图像亚像元定位方法。全变分(total variation,TV)模型是经典的保边缘平滑滤波器,本文将其引入作为预处理,来提高混合像元分解及亚像元定位的精度;本文方法在训练和检验样本的构建过程中,依据空间相关性理论,同时考虑了中心像元及其邻近像元丰度值对亚像元类别归属的影响;在监督分类训练和检验过程中,通过剔除纯净像元来缩减样本数量,在保证算法准确性的同时提高了效率。对真实高光谱遥感数据进行了实验,主观评价和定量分析验证了本文方法的有效性。
[Abstract]:The support vector machine based on support vector machine is proposed. SVM is a subpixel localization method for hyperspectral remote sensing images. The total variation / TVV model is a classical edge-preserving smoothing filter, which is introduced as preprocessing in this paper. To improve the accuracy of mixed pixel decomposition and sub-pixel location; In this paper, in the process of constructing training and testing samples, according to the spatial correlation theory, the effects of the abundance value of the center pixel and its adjacent pixel on the subpixel classification are considered. In the process of supervised classification training and testing, we reduce the number of samples by eliminating pure pixels, and improve the efficiency while ensuring the accuracy of the algorithm. Experiments on real hyperspectral remote sensing data are carried out. Subjective evaluation and quantitative analysis verify the effectiveness of this method.
【作者单位】：中国地质大学地球物理与空间信息学院;湖北省地球内部多尺度成像重点实验室;
【基金】：国家自然科学基金(61271408)~~
【分类号】：TP751
【正文快照】： 混合像元普遍存在于遥感图像中,并且对解译带来了困难。传统分类技术会导致信息丢失,且分类精度难以满足要求。针对混合像元分解无法确定地物空间分布的问题,Atkinson提出了亚像元定位(sub-pixel mapping,SPM)技术[1]。国内外已有不少亚像元定位方法,Mertens提出了利用遗传算

【相似文献】