Using new independent component analysis (ICA) based spectral index to extract and map built-ups of India's sacred district ‘Mathura’

Abstract

Feature extraction through remotely sensed imageries is tricky, especially from images with medium to fine resolution. That's why there is always the scope for new and distinct studies in this field. Various techniques of image processing still need to use for satellite image processing. Some methods are explicitly dedicated to a particular type of remote sensing images, like independent component analysis (ICA), one of the dimensionality reduction methods. ICA has been used in remote sensing for a long but primarily for feature extraction through microwave remote sensing imageries and hyperspectral images. Like principal component analysis (PCA), ICA also has several applications, yet unlike PCA, ICA is not much used for multispectral satellite images. ICA uses the higher order statistics to lessen the spectral mixing and enhancing the precision of feature extraction. Therefore, this study uses ICA but in a different manner, i.e., in the spectral ratio to extract the built-up pattern through the multispectral remotely sensed image for urban mapping of India's sacred Mathura district.