OBJECT IDENTIFICATION USING TEXTURE BASED CLASSIFICATION

Abstract

Identification of different type of objects present on the earth surface is an important aspect in remote sensing application. Depending upon the applications different type of objects can be identified. Traditional land use mapping by visual image-interpretation is expensive, time consuming and often subjective. Researchers have been searching for automatic or semi-automatic approaches for many years. For decades, large-scale aerial photos have been employed to extract building for mapping application. With the successively launching of high-resolution commercial satellites (e. g. IKONOS and QuickBird, WorldView-2), high-resolution satellite imagery has been shown to be a cost-effective alternative to aerial photography in many applications. The overall objective of this research is the development of semiautomatic approach for object identification based on high resolution WorldView-2 image and development of methods which incorporate texture information of objects. This research utilizes object based image analysis (OBIA) approach for a Roorkee region using WorldView-2 imagery, to identify trees and buildings. The methodology involves the texture information and also ancillary information such as shape, size, color, shadow etc. of objects. The method segments the image pixels into objects and utilizes texture and contextual information of the object rather than only using spectral information relied upon by traditional methods. IMAGINE Objective tool from ERDAS Imagine software were used to define feature models for trees and buildings identification. In addition, the results were compared using visual interpretation, by overlaying vector file on the original subset images. Overall, the use of OBIA based on texture information holds great promise to identify and extract individual trees and buildings in Roorkee area. Keywords: Object based image analysis, WorldView-2,