2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference


Title
Stripping assessment of asphalt coating using k-means clustering and support vector machines

Return to Session

Author(s)
Dr. Ehsan Rezazadeh Azar, Lakehead University (Presenter)
Mr. Ashkan Sahari Moghaddam, Lakehead University
Dr. Yolibeth Mejias, Ministry of Transportation Ontario
Mrs. Heather Bell, Ministry of Transportation Ontario
Abstract

Stripping of the asphalt coating is a major moisture-related damage in hot mix asphalt pavements, which deteriorates the bond between the asphalt cement and aggregate particles. This issue could initiate many forms of asphalt pavement distresses, namely ravelling. Static immersion is a common testing method to assess the stripping of asphalt cement cover from the aggregate particles in a submerged condition, but since this assessment depends on the visual judgment of technicians, its accuracy and reproducibility have been disputed by professionals and research community. Image processing and machine learning methods have proven to be reliable tools and have the potential to provide consistent and accurate results in this test. This paper introduces a computer vision-based system to estimate the stripping of test samples processed in the static immersion test. This system employs series of image processing methods to enhance the lighting of the images and to correct specular highlights. Then the pixels on the enhanced images are segmented using the k-means clustering algorithm, and the resulted clusters are classified using linear support vector machines algorithm to determine the number of pixels belonging to the coated and uncoated areas. A set of experiment was carried out to evaluate the performance of this system, in which the machine-measured results did not have a significant difference with the manual assessments with a mean difference of 4.9%.