New multispectral images classification method based on MSR and Skewness implementing on various sensor scenes

Abstract

A new features extraction approach is presented based on mathematical form the modify soil ratio (MSR) and skewness for numerous environmental studies. This approach is involved the investigate on the separation of features using frequency band combination by ratio to estimate the quantity of these features, and it is exhibited a particular aspect to determine the shape of features according to the position of brightness values in a digital scenes, especially when the utilizing the skewness. In this research, the marginal probability density function G(MSR) derivation for the MSR index is corrected, that mentioned in several sources including the source (Aim et al.). This index can be used on original input features space for three different scenes, and then implemented the marginal probability density function of MSR values to stretch the histograms of MSR images without any processing. Skewness is proposed on MSR images and combined with multispectral bands of original scene for land cover classification. This is a new method for extensively observing the types of features and its changes. The Hyperion data were utilized in this work; because they contain abundant details information for distinguish the different types of features.