Toward Optimum Fusion of Thermal Hyperspectral and Visible Images in Classification of Urban Area

Abstract

Recently, classification of urban area based on multi-sensor fusion has been widely investigated. In this paper, the potential of using visible (<small>VIS</small>) and thermal infrared (<small>TIR</small>) hyperspectral images fusion for classification of urban area is evaluated. For this purpose, comprehensive spatial-spectral feature space is generated which includes vegetation index, differential morphological profile (<small>DMP</small>), attribute profile (<small>AP</small>), texture, geostatistical features, structural feature set (<small>SFS</small>) and local statistical descriptors from both datasets in addition to original datasets. Although Support Vector Machine (<small>SVM</small>) is an appropriate tool in the classification of high dimensional feature space, its performance is significantly affected by its parameters and feature space. Cuckoo search (<small>CS</small>) optimization algorithm with mixed binary-continuous coding is proposed for feature selection and <small>SVM</small> parameter determination simultaneously. Moreover, the significance of each selected feature category in the classification of a specific object is verified. Accuracy assessment on two subsets shows that stacking of <small>VIS</small> and <small>TIR</small> bands can improve the classification performance to 87 percent and 82 percent for two subsets, compare to <small>VIS</small> image (72 percent and 80 percent) and <small>TIR</small> image (50 percent and 56 percent). However, the optimum results obtained based on the proposed method which gains 94 percent and 92 percent. Furthermore, results show that using <small>TIR</small> beside <small>VIS</small> image improves classification accuracy of roads and buildings in urban area.