摘要
为了解决在应用传统高光谱地物识别方法时,由于吸收峰个数不同,造成的光谱匹配误差较大的问题,采用了一种基于高光谱吸收峰特征的选择方法,根据选择后的吸收峰特征进行光谱曲线匹配。该算法首先对高光谱曲线进行包络线消除并提取光谱特征参量矩阵,然后根据标准特征参量矩阵与待测特征参量矩阵的每个向量的余弦距离-欧氏距离来逐一寻找吸收峰的匹配向量,之后根据选择的吸收峰特征参量矩阵进行了理论分析和实验验证。结果表明,该算法可以搜寻到最佳的特征参量向量,从而实现吸收峰的选择,用选择后的吸收峰的特征参量矩阵进行高光谱匹配。这一结果对降低匹配的误差是有帮助的。
When adopting traditional hyperspectral ground objects identification method, the error of spectral matching became big because of the difference of absorption peak number. In order to solve the problem, an optional algorithm based on hyperspectral absorption peak characteristics was brought out, and then spectral matching according to the selected absorption was carried out. At beginning, the continuum removal in spectral curve and the extraction of spectral characteristic parameters matrix were made. And then the matching vertor of absorption peak was searched gradually according to the cosine distance-Euclidian distance of every vertor from the standard characteristic parameter matrix and the to-be-detected characteristic parameter matrix. After theoretical analysis and experimental verification of the selected absorption peak characteristic parameters matrix, the results show that this algorithm can get the best characteristic parameters vector, to realize the selection of absorption peak and make the hyperspectra matching with the selected absorption peak characteristic parameters matrix. The study is helpful for the decrease of the error of spectral matching.
出处
《激光技术》
CAS
CSCD
北大核心
2016年第6期848-852,共5页
Laser Technology
基金
国家自然科学基金资助项目(61375011)
浙江省自然科学基金资助项目(LY13F030015)
关键词
光谱学
光谱匹配
吸收峰选择
向量距离
spectroscopy
spectral matching
absorption peak selection
vector distance
