Automated pavement surface distress evaluation would be a useful component of computerised pavement management systems. Methods are described for applying computer simulations of pavement distress appearance to the design of those evaluation systems. By generalising the results of crack visibility calculations under varying conditions of lighting, a number of conclusions about automated evaluation systems can be drawn, including the fact that there will always be some distress that escapes detection but that the detection probability is calculable. Lighting is important in determining which types of distress and what fraction of distress will evade detection. Differences between controlled and natural lighting, between ambient or omnidirectional lighting and directed lighting, and between single versus multiple light sources are discussed. Finally, there is consideration of how detailed examination of modelling results can be used to optimise the design of lighting, data acquisition hardware, and image-processing software. Because the conclusions depend strongly on the assumptions made in the work, a detailed listing of the assumptions is presented along with suggestions for further testing.
Abstract