Higher Order Statistics for Multispectral Satellite Data

Abstract

Satellite Data concisely convey information about positions, sizes and interrelationships between objects. The satellite image losses information due to lack of Acquisition capability of sensor and atmosphere's effect. It is very difficult to extract useful information at intensity level with low SNR, non wavelet segmented schemes losing high frequency contact with results texture is blurred several preprocesses are applied to make textual image clear and segmentation. Unsatisfied results due with lack of directionality with DWT, Here we can implement advance image processing technique for improving texture based features to multispectral satellite image, find discrepancy distribution of observed and normal region using Higher order statistical methods(HOS) like skewness, Kurtosis. The shape of the distribution of intensity levels are examined by HOG. For improving the visualization quality we examine features based on edges, lines and their gradients using Curvelet and Histogram of oriented Gradient (HOG), intensity distribution using Higher order Statistics (HOS).