Automatic Road Extraction from Airborne LiDAR Data in Urban Area

Automatic road extraction from urban area is a challenging issue because of highcomplexity. An approach to achieve automated road extraction from Airborne LaserScanning (ALS) data is presented. The proposed approach consists of two steps. In thefirst step, Support Vector Machine (SVM) algorithm was applied on intensity anddistance data merely to classify data into two groups of road and non road categories. Inthe second step, the road category was improved by removing unwanted non-road pixelsusing morphological operation. Due to the lack of aerial photos of the study area,reference map prepared by manual digitizing of Google Earth images. Throughcomparing the achieved results with ground truth road map as reference data, theperformance measures such as completeness, correctness, and quality were calculated.Results of the evaluation showed the value completeness, correctness, and quality were84.01%, 79.92%, and 69.37% respectively. These measures prove that the proposedapproach detects and extract roads in a robust manner. The results of this study show thatmorphological operators play an important role in improving the accuracy and quality ofthe results.

کلیدواژه ها:

Airborne LiDAR ، SVM ، Road extraction ، Morphological operator