An enhanced approach for detection of boulders on natural hillsides using image processing and LiDAR technologies

Boulders sitting on natural hillsides can pose a hazard to life and facilities downslope if they become unstable. To accurately assess boulder fall hazards of any hillside, it is important to map the locations and distribution of these boulders. Without such information, stability analysis and appropriate mitigation measures cannot be carried out. However, some of these boulders are scattered on steep natural hillsides and often covered by thick vegetation. Comprehensive field mapping of these boulders is time-consuming and labour-intensive, sometimes incomplete and may expose field personnel to an unsafe working environment. To overcome these constraints and reduce the amount of field mapping, remote sensing technology is considered a useful and effective alternative for boulder mapping. The Hong Kong Polytechnic University (HKPolyU) and the Geotechnical Engineering Office (GEO), Civil Engineering and Development Department of the Government of the Hong Kong Special Administrative Region have previously attempted to employ image processing to map boulders on natural hillsides and achieved a satisfactory level of accuracy. To enhance the previous approach, so as to increase the reliability and accuracy in boulder identification, a study adopting both image processing and airborne LiDAR technologies has recently been completed. In this study, the data from image processing and airborne LiDAR were extracted and processed using tailor-made algorithms to generate boulder polygons with their corresponding attributes such as dimensions, location and slope gradient. The results of this study indicate that this enhanced approach has significantly improved the reliability and accuracy of boulder identification. As only little fieldwork is required in this approach, it can avoid undesirable vegetation clearance required for provision of field access and reduce the risk to field personnel working on steep natural hillsides. This paper outlines the principles, assumptions, and details of the workflow for the enhanced approach for boulder detection on natural hillsides.