基于谱线形状与信息量差异的高光谱解混NMF初始化方法
发布时间：2019-06-18 09:31
【摘要】：在高光谱像元解混应用中,好的端元光谱矩阵初始化方法对于提高盲信号分解精度具有重要意义。针对空间分辨率较高的高光谱数据,提出了一种新的面向非负矩阵分解(non-negative matrix factorization,NMF)的初始化方法。该方法通过计算像元在谱线形状和信息量差异等方面的参数,利用像元谱线峭度、KL散度和光谱角等参量,从众多混合像元中识别出纯像元;并分辨出不同类型纯像元(或类纯像元)之间的差别,从中选择最适合代表每一类型端元的纯像元(或类纯像元)作为算法的初值像元,完成端元矩阵的初始化。将此方法分别用于模拟数据和真实数据的实验结果表明,该方法能够明显提高高光谱混合数据的NMF精度,相比其他常用初始化方法具有更好的效果。
[Abstract]:In the application of high-spectral image element, a good end-element spectral matrix initialization method is of great significance to improve the accuracy of blind signal decomposition. A new method for initializing non-negative matrix factorization (NMF) is proposed for high spectral data with higher spatial resolution. The method comprises the following steps of: calculating a parameter of an image element in a spectral line shape and an information quantity difference and the like, identifying a pure image element from a plurality of mixed image elements by using a parameter such as a spectral line similarity, a KL divergence angle and a spectral angle, and distinguishing the difference between different types of pure image elements (or quasi-pure image elements), A pure image element (or quasi-image element), which is most suitable for representing each type of end element, is selected as the initial value image element of the algorithm, and the initialization of the end element matrix is completed. The experimental results show that the method can obviously improve the NMF accuracy of the high-spectral mixed data, and has a better effect than the other common initialization methods.
【作者单位】： 南阳理工学院数学与统计学院;南阳理工学院经济管理学院;
【基金】：河南省高等学校重点科研项目“Smith正规型在有限域上有理点个数中的应用”(编号:17A110010)资助
【分类号】：TP391.41
本文编号：2501370
[Abstract]:In the application of high-spectral image element, a good end-element spectral matrix initialization method is of great significance to improve the accuracy of blind signal decomposition. A new method for initializing non-negative matrix factorization (NMF) is proposed for high spectral data with higher spatial resolution. The method comprises the following steps of: calculating a parameter of an image element in a spectral line shape and an information quantity difference and the like, identifying a pure image element from a plurality of mixed image elements by using a parameter such as a spectral line similarity, a KL divergence angle and a spectral angle, and distinguishing the difference between different types of pure image elements (or quasi-pure image elements), A pure image element (or quasi-image element), which is most suitable for representing each type of end element, is selected as the initial value image element of the algorithm, and the initialization of the end element matrix is completed. The experimental results show that the method can obviously improve the NMF accuracy of the high-spectral mixed data, and has a better effect than the other common initialization methods.
【作者单位】： 南阳理工学院数学与统计学院;南阳理工学院经济管理学院;
【基金】：河南省高等学校重点科研项目“Smith正规型在有限域上有理点个数中的应用”(编号:17A110010)资助
【分类号】：TP391.41
【相似文献】
相关期刊论文 前5条
1 郑金彬;欧阳利军;;基于密度等高线的原型初始化方法[J];南华大学学报(自然科学版);2007年01期
2 汪西莉;刘芳;焦李成;;一种小波域HMT模型参数初始化方法[J];计算机科学;2003年01期
3 曹彩凤;面向对象的C++程序设计中各类对象初始化方法[J];计算机应用与软件;2003年03期
4 杨圣云;袁德辉;赖国明;;一种新的聚类初始化方法[J];计算机应用与软件;2007年08期
5 孙泽强;谢红薇;郝晓燕;;NMF模型在挖掘基因功能关系中的研究与应用[J];计算机工程与设计;2014年04期
相关硕士学位论文 前1条
1 李靖;基于小波变换和NMF的数字图像水印算法[D];湖北工业大学;2011年
，本文编号：2501370
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2501370.html
