关键词: 高光谱影像分类, 改进二进制蚁群算法, 波段选择, 光谱特征, 纹理特征

Abstract: An Improved Binary Ant Colony Optimization（IBACO） algorithm is proposed, which combined with the initial heuristic information of Genetic Algorithm（GA） and the global optimization ability of Binary Ant Colony Optimization（BACO） algorithm to overcome the high dimensionality of hyperspectral image. In the proposed method, the results of GA are utilized as the initial heuristic information of BACO, and the ant path selection mechanism is improved to enhance the global optimization ability. On the other hand, texture feature is utilized to make full use of the spectral and spatial information, and the combination of spectral and texture can obtain higher classification accuracy. Experimental results illustrate that the search ability of IBACO is better than GA, ACO, BACO. Meanwhile, the classification accuracy by using the proposed technique has reached 95.63%. In all, the proposed method can effectively enhance the classification efficiency and classification accuracy.

Key words: hyperspectral image classification, Improved Binary Ant Colony Optimization（IBACO） algorithm, band selection, spectral feature, texture feature