Multiscale and efficient recognition of land cover by using both decomposition and deep visual feature analysis

Abstract

As we know that when we use the colour bands that are pure in nature it will help some of the public authorities to update the land information. It will be a sensing platform so it will take less time and the cost we will be spending will not be much heavy, in that manner aircraft system will be used that should be unmanned. Here we will be using a method for optical aerial imagery that will come under land use classification with decision tree. In the beginning the ownership parcel map will come under play for taking the land cover information through maximum likelihood classifier. Then, to make the relation between land usage and land coverage we will generate the decision tree. A well structured parcel map will be produced by the merits of geographical characteristics of parcels.Hence the aftereffects of the spatial resolution under the classification approach will be discussed completely by sampling them again from the 20 then 50 then 100cm. This land use can be widely used in landscape and urban planning because of the flexible classification method. Using aerial images the object oriented approach for image based analysis is used for urban land cover mapping.