摘要
高光谱图像分类是许多应用的第一步,也是极其重要的一步。针对目前分类方法存在误分现象,尤其是在地物边缘附近区域,以及现有空谱联合分类方法计算复杂度高的问题,提出一种基于本征图像分解以及导向滤波的高光谱图像空谱联合分类方法:利用AP聚类进行波段选择,提高计算效率;利用基于局部稀疏约束的本征图像分解方法进行高光谱本征图像分解,获取反射率本征图;利用导向滤波器对初始分类结果进行优化。实验结果表明:文章提出的空谱联合分类方法在分类精度与计算时间方面优势明显。
As a kind of high-resolution remote sensing, hyperspectral remote sensing has a wide range of application prospects because it can provide rich spectral information. Hyperspectral image classification is the first and most important step in many applications. There is a misclassification phenomenon in the current classification method, especially in the vicinity of the edge of the object, and the existing classification method with spatial-spectrum combination has high computational complexity. A spectral-spatial classification method for hyperspectral imagery based on intrinsic image decomposition and guided filtering is proposed. Affinity propagation (AP) clustering is used for band selection to improve computational efficiency. The method of hyperspectral intrinsic image decomposition based on local sparseness is performed to obtain the reflectance intrinsic image. The guided filter is utilized to optimize the initial classification results. The experimental results show that the proposed method has obvious advantages in classification accuracy and computation time.
作者
任智伟
吴玲达
REN Zhiwei;WU Lingda(Jiuquan Satellite Launch Center, Jiuquan 732750, China;Space Engineering University, Beijing 101416, China)
出处
《航天返回与遥感》
CSCD
2019年第3期111-120,共10页
Spacecraft Recovery & Remote Sensing
关键词
光谱学
高光谱图像
本征图像分解
局部稀疏约束
空谱联合
导向滤波
遥感技术应用
spectroscopy
hyperspectral image
intrinsic image decomposition
local sparseness
spectral-spatial
guided filter
remote sensing technology application
