Sevcik’s Fractal Based Dimensionality Reduction of Hyper-Spectral Remote Sensing Data

Recent advancements in Remote sensing has led the way for the development of Hyper spectral sensors. Hyper spectral remote sensing is a new technology that is currently being used by researchers and scientists with regard to the detection and identification of mineral, vegetation, man-made materials and other features. These sensors are far more superior as they collect information in a very narrow contiguous wavelength interval. Hyper spectral data carries information of the features in ten to thousand bands. This wealth of data is hard to exploit, as it needs very high computational complexity to store and to process this data. Hyper spectral data are also highly correlated, so special attention is needed to reduce the dimensions of hyper spectral data. The spectral reflectance of any pixel depends on the characteristics of a land cover class. The fractal dimension of the spectral reflectance curve (SRC) of any pixel can thus be calculated. Based on this, fractal based method can thus be employed to reduce the dimensionality of hyper spectral remote sensing data. The fractal dimension of SRC is calculated by sevcik’s method. As a case study, AVIRIS 224 band data is used and processed in MATLAB 7.8.0. The study shows that the dimensionality of hyper spectral remote sensing data can be reduced with minimal data loss.