Abstract: Hyperspectral images are used in wide range of applications in remote sensing domain such as agriculture, meteorology, mineralogy and surveillance. Hyperspectral images are acquired at vast electromagnetic spectrum of about 400-2500 nm which provide increased sensitivity and high ability to distinguish the objects. Due to its high spectral resolution these images exhibit the spectral signatures unique to specific objects, minerals and vegetation. In this research, we envisages a new spectral signature based classifier in which each pixel is assigned to the exact classes based on the spectral signatures or similar spectral statistical characteristics for the effective classification of hyperspectral images. HyDICE urban image dataset is used to assess the performance characteristics of the proposed algorithm and also compared with conventional classifiers of k-means and fuzzy c-means methods. The experimental results shows that the proposed spectral signature based classification method outperforms the existing classification methods.
S. Chidambaram and A. Sumathi, 2016. A Novel Spectral Signature Based Classification Approach for Airborne and Spaceborne Hyperspectral Imagery. Asian Journal of Information Technology, 15: 4926-4933.