A comparison of road network extraction from High Resolution Images

Study related to high resolution remote sensing (RS) images has been applicable to many areas that are beneficial to us for information related to territory, public, weather and agriculture, the main aim of RS applications is extract appropriate information that will help us to draw meaningful conclusions. Road network extraction is an important and challenging research field, road networks are essential for humans as they provide transportation and other support systems. Distinct factors like sensor, weather, resolution and light etc., can affect the road features from a RS image which imposes the problems in road network extraction. This paper presents a comprehensive analysis of various aspects of road network extraction from RS images like road features, problems in road network extraction and finally different road network methods are classified on the basis of local and global features, automation and algorithm used. Different road extraction techniques are compared on the basis of features used in extraction, number and type of data (aerial, hyper-spectral, remote sensing, urban and semi-urban) used, performance on the basis of different qualitative parameters and the advantage and disadvantages of different methods are discussed. The comparative analysis of road extraction methods is presented emphatically and it is observable that in order to obtain precise road network from RS images only one type of feature is not sufficient. Hence, multiple road features should be combined together but it depends on type of application and data. Road network extraction from RS image is still remains a challenging and important research field.