矩阵因子和信息散度融合的高光谱波段选择方法

A New Band Selection Method for Hyperspectral Images Based on Matrix Mode and Spectral Information Divergence

高光谱数据的数据量大、冗余度高,影响了影像分类的精度和速度。为解决这一问题,在基于矩阵模式的高光谱波段选择方法的基础上,提出了一种基于矩阵因子和光谱信息散度融合的波段选择方法。首先,用光谱信息散度代替原始用光谱相关系数衡量光谱维的相关性。然后,调整矩阵因子定义方式,根据单一量化指标大小选择最佳波段组合。最后,分别用最佳指数因子、自适应波段选择法、最优波段指数法波段和所提出的算法对AVIRIS高光谱数据进行波段选择试验。结果表明,在选择相同波段数目的情况下,所提出的算法最大化保留了光谱信息并去除了冗余波段,分类精度有明显提高。

Hyperspectral data have the characteristic of enormous data and high redundancy, which affects the accuracy and speed of image classification. In order to solve this problem, this paper proposes an improved method of band selection BSMM-SID,which is based on the Hyperspectral band selection method. Firstly, use spectral information divergence to measure the correlation of spectral dimension, displacing the original method which uses spectral correlation coefficient.Next, adjust the definition of matrix factor. Then, select the best band combination according to the size of a single quantization index. Finally,the band selection test of AVIRIS hyperspectral data is carried out by using OBI, ABS,BSMM and BSMM-SID respectively. The test results show that the proposed method can significantly improve the working efficiency of the classification method after dimension reduction by removing the redundant bands and preserving the maximum spectral information in hyperspectral data.