一种动态调整权值的交叉相关光谱匹配算法

Cross Correlation Spectral Matching Algorithm of Dynamic Adjustment Weights

针对传统算法对于光谱距离特征缺乏描述,并且无法动态调整光谱之间的相似程度这一问题,加入2个光谱特征的评价指标因子:均方根误差与欧氏距离,将二者的权重进行动态调整,得到实验精度在权重系数最优情况下,比基于欧氏距离的匹配方法精度提高了20%以上,且比传统交叉相关光谱匹配方法更加稳定。实验证明,该方法能够较好地改善交叉光谱匹配算法,利用动态的权重系数调整比传统方法更合适于用于高光谱影像的光谱匹配和信息提取。

The traditional matching algorithms lack the description of spectral distance,and you cannot dynamically adjust the degree of similarity between the spectra.To solve this problem,this paper introduced the RMS and Euclidean distance which could be dynamically adjusted as the evaluation factors of spectral characteristics.Finally,the results obtained by analyzing and comparing metrics.Experiments show that the method can improve cross correlation spectral matching algorithm.Dynamic adjustment of the weights is more appropriate than the traditional method for spectral matching and information extraction on hyperspectral image.