Hyperspectral imaging based land cover mapping using data obtained by the Hyperion sensor

This paper presents an analysis of hyperspectral image data corresponding to a strip along the North Eastern region of Sri Lanka, obtained by the Earth Observing (EO-1) satellite's Hyperion sensor. Using hyperspectral imagery in order to map land-cover maps is beneficial in many ways as it could be used as a basis to obtain useful information for natural resource and ecosystem service management, assessing the human induced and natural drivers of changes in land, foliage or water bodies and even in identification of fine details such as the distribution of minerals in an area. In the algorithm proposed in this paper, each pixel was represented as a point in a high dimensional space of which the dimensions represented each band of wavelength. Principal Component Analysis (PCA), Fisher Discriminant Analysis (FDA) and Spectral Clustering were used in a logical sequence, as discussed in this paper, in order to cluster the points in a reduced dimensional space. The pixels belonging to each cluster were labeled under ‘soil’, ‘foliage’ or ‘water bodies’, with the aid of the k-means algorithm and the hyperspectral data of the training set obtained with the aid of Google Maps. Upon validation it was observed that the procedure employed is an effective and promising method of classifying a semi supervised hyperspectral dataset.