摘要
在研究了非负矩阵分解(NMF)用于遥感图像融合技术的基础上,改进了一种基于光谱复原的NMF遥感图像融合算法,首先利用矩阵非负性分解求取图像的基向量矩阵,然后综合考虑图像融合后的光谱特性和波形变化情况,对基向量矩阵进行直方图规则化处理,最后得到融合后的图像。实验结果表明,该方法具有较好保持原始图像的光谱信息和空间信息的优点。
After researching on the application of the Non-negative Matrix Factorization （NMF） in remote sensing image fusion, an improved method based on spectrum reconstruction is brought out in this paper. Firstly, image matrix is decomposed with the prospect of non-negative to get the basic vector of the matrix. Then, the basic vector is histogram regularized based on the integration of spectral characteristic and wave variety, and the final result could be gained. The test results show that the method in this paper posses the virtue of preserving spectral and special information of the original image.
出处
《海洋测绘》
2010年第1期41-43,共3页
Hydrographic Surveying and Charting
关键词
遥感影像
图像融合
矩阵分解
光谱复原
remote sensing image
image fusion
matrix factorizafion
spectrum reconstruction
