回顾了传统高光谱图像的特征选择算法所存在的问题,将支持向量机分类器和自适应小生境遗传算法引入到高光谱图像的特征选择上,提出了一种高光谱图像特征选择算法,该算法充分考虑了高光谱图像的波段相关性,同时克服了传统遗传算法的早熟问题和支持向量机的参数选择问题,以微小的代价,大大地压缩了光谱矢量的特征维数,达到令人满意的分类效果。 The existing problems of the feature selection algorithm of traditional hyperspectral images are reviewed. The support vector machine classifier and adaptive niche genetic algorithm are introduced to the feature selection of hyperspectral images, and a hyperspectral image feature selection algorithm is proposed. The algorithm takes full account of the band correlation of hyperspectral images and overcomes the premature problem of traditional genetic algorithms and the parameter selection of support vector machines. At the same time, the algorithm greatly reduces the feature dimension of spectral vector to a satisfactory level The classification effect.