Potential of Low-Cost, Close-Range Photogrammetry Toward Unified Automatic Pavement Distress Surveying

Automatic pavement distress detection and data collection is important for pavement management systems. It is estimated that pavement defects cause damage costing $10 billion/year in the US alone. Despite the importance of image-based distress detection systems, they are still semi-automatic to a great extent. They rely internally on one or more threshold values during processing or may need a pre-processing stage, and the quality is affected by shadows and low or extra illumination among other factors. After years of research, processing still typically relies heavily on global or in-context pixels content analysis. Such systems lack the robust sensor modeling, hence, robust detection and modeling which cannot be achieved directly through 2D image space analysis. The exploitation of arrays of laser profilers for 3D data acquisition is an expensive approach and has limitations for enhancing or replacing image-based output. Alternatively, 3D surfaces can be generated using stereo vision techniques. This research has investigated close range photogrammetry as a robust approach to overcome the above disadvantages. The experimental work is carried out using a non-metric DSLR camera with its built-in flash and natural daylight as sources of illumination. Initial investigations show significant potential for 3D distress detection and modeling with higher spatial precision and a higher level of automation, while retaining 2D color and shading information for data fusion. The output of automatic photogrammetric processing can be further exploited directly in existing automated and semi-automated systems for updating the content, analysis and visualization of pavement management system (PMS) and geographic information systems (GIS).