基于广义判别分析的高光谱影像特征提取

为提高高光谱影像地物识别的精度与速度,采用基于核方法的广义判别分析进行高光谱影像的非线性特征提取.研究了广义判别分析的数学模型、模型求解方法及特征提取过程,并进行了高光谱影像特征提取与分类实验.结果表明:样本点在特征空间中,同类目标大体聚集成团,异类彼此分离,具有良好的紧致性,特征提取结果优于线性判别分析结果.

Hyperspectral image feature extraction based on generalized discriminant analysis

The generalized discriminant analysis based on kernel method was used to extract nonlinear feature of hyperspectral image to improve the precision and speech of hyperspectral image classification.The mathematical model for the general discriminant analysis and its solving method were studied,and then the feature extraction process was given.Tests on hyperspectral image feature extraction and classification show that the samples are in the feature space,and samples of the same class are close to each other,and the samples of the different classes are far away.The method has good compactness,whose result of feature extraction is superior to that of linear discriminant analysis.