摘要
提出了基于互信息加权的最小二乘算法丰度反演,选择互信息矩阵作为加权矩阵,从熵的角度反映了不同波段间的相关性.同时,在丰度反演过程中应用波段选择技术,降低了数据处理的复杂度.分析实验仿真结果,与传统的最小二乘算法和已有的加权最小二乘丰度反演算法相比,获得了更精确的丰度信息,反演效果得到提升,验证了该算法的可行性.
In order to highlight the distinctness between the bands and obtain more accurate abundance of mixed pixels, the least squares error algorithm is used, which is based on weighted matrix for the abundance inversion. Abundance inversion based on mutual information-weighted least squares error algorithm is presented, mutual information from the perspective of entropy to reflect the correlation between different bands. Band selection technology is adopted in abundance inversion to reduce the complexity of data processing. Compared with the existing weighted matrix and traditional least squares error problem, the analysis of the experimental result shows the feasibility of this algorithm.
出处
《沈阳大学学报（自然科学版）》
CAS
2014年第1期45-49,共5页
Journal of Shenyang University：Natural Science
基金
国家自然科学基金资助项目(61077079)
教育部博士点计划基金资助项目(20102304110013)
黑龙江省自然科学基金重点资助项目(ZD201216)
关键词
高光谱解混
丰度反演
最小二乘算法
互信息
波段选择
hyperspectral unmixing
abundance inversion
least squares error algorithm
mutual information
band selection
