Detection of Oil Tank from High Resolution Remote Sensing Images using Morphological and Statistical Tools

Abstract

Oil tank is an important target and automatic detection of the target is an open research issue in satellite based high resolution imagery. This could be used for disaster screening, oil outflow, etc. A new methodology has been proposed for consistent and precise automatic oil tank detection from such panchromatic images. The proposed methodology uses both spatial and spectral properties domain knowledge regarding the character of targets in the sight. Multiple steps are required for detection of the target in the methodology – 1) enhancement technique using directional morphology, 2) multi-seed based clustering procedure using internal gray variance (IGV), 3) binarization and thinning operations, 4) circular shape detection by Hough transform, 5) MST based special relational grouping operation and 6) supervised minimum distance classifier for oil tank detection. IKONOS and Quickbird satellite images are used for testing the proposed algorithm. The outcomes show that the projected methodology in this paper is both precise and competent.