リモートセンシングとGISのジャーナル

概要

Automatic Urban Road Extraction from High Resolution Satellite Data Using Object Based Image Analysis: A Fuzzy Classification Approach

Yadav DP*, Nagarajan K, Pande H, Tiwari P, Narawade R

In the present world with a severe economic crisis, time and cost are the most crucial factors for any project. Civil engineering is a field of site execution where activities need an ample of time for its completion due to many factors and any contagious pandemic situation like Covid-19 makes difficult to perform the onsite operations. Hence adopting satellite data for planning purpose where required features can be automatically extracted and analyzed in any GIS software, is significant. This paper aims at the extraction of road features from high-resolution satellite data employing fuzzy classification technique. Worldview-2 satellite data of Gandhi nagar, the capital city of Gujarat, India having 0.5 m panchromatic and 2 m multispectral resolution, is used. Image fusion is carried out by using the bilinear sampling technique of Principal Component Analysis to obtain 0.5 m multiresolution pan-sharpened satellite image. Road feature is extracted by performing multiresolution image segmentation and developing a rule set for classification by adopting the object-based image analysis method. Its accuracy assessment attains the completeness of 71.65%, correctness of 70.33% and quality of 59.98%. This method provides a rapid novel approach for feature extraction with comparatively less data availability as no sun illumination divergence or thematic knowledge or altitude information are used, leading to pandemic suitable remote accessibility and cost-effective approach.