Machine Vision for Road Pavement Applications Bitumen Coverage and Grain Size Estimation

Abstract: In this thesis some research questions regarding durability and quality of roads has been investigated. The questions are analyzed from an image analysis point of view and aims to be a complement to existing methods for analyzing asphalt. One important factor for the durability of the asphalt layer on roads is the affinity between the stones in the asphalt and the binder that holds the stones together, called bitumen. One step in testing the affinity is to manually estimate the degree of bitumen coverage after the stones covered in bitumen has been washed for a while. The goal with the first two papers is to replace this manual estimation by image analysis methods. The first paper deals with the easier problem where there is a clear color difference between the stones and the bitumen. By using reference images to get information of the typical stone and bitumen color and a graph-cut algorithm we get result that seems to be close to the real degree of bitumen coverage. In the second paper we instead look at the problem with darker stones. In this case we cannot see a clear color difference between the stones and the bitumen. Instead we notice that bitumen and stones reflect light in different ways and take multiple images with lighting from different directions. The degree of bitumen coverage is then estimated by detecting specular reflections in the images. Another quality control of asphalt is to estimate the size distribution in an asphalt sample and see if it corresponds to the recipe for the asphalt. This is investigated in the third paper, where slices of the asphalt are analyzed. The analysis consists of segmenting the stones individually so that the size of all grains can later be estimated.