A New Approach to Semi-automated Road Extraction from Orthorectified Radar Imagery

Extracting road features from orthorectified radar images is usually a difficult task due to speckle noise and radiometric obstructions alongside the road. The common approach to this task is therefore to extract road features manually, which is very time-consuming and labour intensive. In this paper, a semi-automated procedure is proposed which integrates the common manual tool with a new automated functionality. The automated road feature extraction functionality takes advantage of the information intrinsic to the radar image and also of the knowledge of road design specifications. Preliminary results have shown the proposed approach improves both user productivity and product quality.