Road Extraction Techniques from Remote Sensing Images: A Review

Abstract

AbstractRoad detection from remotely sensed images is a fundamental task in the geographic information system. On account of applications like urban management, traffic control, and map updating, road extraction from remote sensing images has significant research importance in recent times. Road extraction from satellite images is a crucial task as these images are noisy and contain lots of information. So it becomes difficult to process large amount of data. The important parameters for road detection are road features and its corresponding classification methods. These parameters decide the performance accuracy of the road extraction system. The systematic analysis of existing road detection techniques is elaborated in three important sections: different features of road, supervised, and unsupervised classification techniques. The main objective of this comprehensive survey is to render the analysis of different classification methods like mathematical morphology, SVM, CNN, etc. By using multiple features of the road, the system performance can be improved.KeywordsRemotely sensed imagesMathematical morphologySVMCNN