Pavement crack classification based on tensor factorization

Abstract

This paper presents tensor factorization as a means of classifying cracks in pavements. Several crack classification algorithms exist and they are mostly based on other machine learning methods. These may come with their own problems which may be classification errors and long processing times. Tensors are multidimensional arrays. The nature of tensors enables the analysis of the images to be carried out in a 3D space which ensures a more robust and accurate analysis tool. The levels of accuracy obtained after using the algorithm implies that crack classification based on tensor factorization is one that can be successfully employed by state agencies nationwide and around the world which use digital image processing systems as part of their pavement management programs.