A simple framework of segmentation for hyperspectral images using clustering techniques

Hyperspectral imaging has been gaining popularity and has been gradually applied to many fields besides remote sensing. Hyperspectral data provides unique information about material classification and reflectance analysis in general. Although hyperspectral images provide abundant information about bands, their high dimensionality also substantially increases the computational burden. However, due to the high dimensionality of the data, both human observers as well as computers, have difficulty interpreting this wealth of information. An important task in hyperspectral data processing is to segment of the spectral image without losing any valuable details. In this paper, we propose a simple framework of segmentation for hyperspectral images using clustering techniques. The proposed framework consists of three main steps. First, dimensional reduction was used to reduce the dimensionality and make it convenient for the subsequent processing steps for hyperspectral images. Secondly, band selection was used for attribute selection in hyperspectral images. Finally, segmentation using clustering technique is employed to automatically segment out of the interested regions in hyperspectral images. The results from the tests confirm the effectiveness of the proposed method in segmentation using our framework for hyperspectral images.