基于萤火虫算法的高光谱遥感波段选择方法

高光谱图像在遥感领域中的应用越来越广泛,但由于自身的高数据维、波段间的高冗余度等特性给图像处理带来了一定困难,针对这个问题,提出一种基于类间可分性准则的改进萤火虫仿生算法,进行高光谱遥感波段选择。在分析萤火虫算法机理的基础上,阐述了利用该算法进行高光谱波段选择的思路,并构造波段相似性矩阵,选择欧氏距离、JM距离、光谱信息散度和离散度作为可分性准则来设置目标函数,根据目标函数值的优劣选择优势波段。最后,使用HYDICE Washington DC Mall和 HyMap Purdue Campus两个高光谱遥感影像数据进行实验验证,并利用支持向量机分类器对最佳波段组合进行精度评价,证明该算法的可行性和有效性。


A Novel Hyperspectral Band Selection MethodUsingImproved Firefly Algorithm

Hyperspectral image has increasingly wide applications in remote sensing field.However,its own high dimension data and high redundant in inter|band takes certain difficulties.For this issue,the paper put forward a novel algorithm by improving the firefly algorithm based on between|class separability criteria to precede band selection.Specifically,motivated by firefly algorithm,the idea and framework using bio|inspired algorithm for hyperspectral band selection are described,similarity matrix in inter|band is designed,Euclidean distance,J|M distance,spectral information divergence as between|class separability criterion are used for objective function,and the discriminant bands based on the merits of target value are chosen.In addition,the experiments and performance assessment were conducted by HYDICE Washington DC Mall and HyMap Purdue Campus data.The experiment results have proved the promising ability of the proposed method for hyperspectral band selection.