基于核方法的高光谱遥感图像混合像元分解

发布时间：2018-05-22 18:43

本文选题：高光谱遥感 + 核方法　；参考：《国土资源遥感》2017年01期

【摘要】：为了提高高光谱遥感图像混合像元分解的精度,提出基于核方法的高光谱线性解混算法。采用正交子空间投影(orthogonal subspace projection,OSP)算子、最小二乘正交子空间投影(least squares OSP,LSOSP)算子、非负约束最小二乘(nonnegative constrained least-squares,NCLS)算子和全约束最小二乘(fully constrained least-squares,FCLS)算子等方法分别构建核正交子空间投影(kernel OSP,KOSP)、核最小二乘正交子空间投影(kernel LSOSP,KLSOSP)、核非负约束最小二乘(kernel NCLS,KNCLS)和核全约束最小二乘(kernel FCLS,KFCLS)高光谱图像混合像元解混模型;对美国内华达州CUPRITE矿区AVIRIS数据进行KLSOSP,KNCLS和KFCLS与LSOSP,NCLS和FCLS丰度反演对比实验。结果表明:对于混合像元广泛存在的高光谱遥感图像来说,基于核方法的KLSOSP,KNCLS和KFCLS的解混精度优于LSOSP,NCLS和FCLS,其中又以KFCLS解混的精度最高;附加约束条件有利于提高丰度反演的精度。
[Abstract]:In order to improve the resolution accuracy of hyperspectral remote sensing images, a kernel based hyperspectral linear de-mixing algorithm is proposed. The orthogonal subspace projection OSP) operator and the least square orthogonal subspace projection least squares OSPP operator are used. Non-negative constrained least-squares (NCLS) operator and fully constrained least-FCLS operator (NCLS) are used to construct kernel orthogonal subspace projection kernel OSPK KLSOSPS, kernel least squares orthogonal subspace projection kernel LSOSPKLSOSPS, kernel non-negative constrained least squares kernel NCLS KNCLS) and kernel orthogonal subspace projection kernel LSOSPKLSOSPS, respectively. Fully constrained least square kernel KFCLS) hyperspectral image mixed pixel demultiplexing model; The AVIRIS data of CUPRITE mining area in Nevada, USA are compared with KLSOSPNCLS and KFCLS and LSOSPNCLS and FCLS abundance inversion experiments. The results show that for hyperspectral remote sensing images where mixed pixels are widespread, the resolution of KLSOSPNCLS and KFCLS based on kernel method is better than that of LSOSPNCLS and FCLS, in which KFCLS is the most accurate, and the additional constraint condition is helpful to improve the accuracy of abundance inversion.
【作者单位】：重庆交通大学土木工程学院测绘系;成都理工大学地学空间信息技术国土资源部重点实验室;重庆市地理信息中心;
【分类号】：TP751.1

【相似文献】