Improving images of asphalt concrete pavements based on segmentation methods
Abstract
Improving images of asphalt concrete pavements based on segmentation methods
Incoming article date: 01.05.2023To assess the quality of the road surface, there are many systems that work on the basis of specific algorithms, including image segmentation methods. Time complexity and classification accuracy are two key indicators when evaluating the effectiveness of a particular algorithm. In this article, the following image segmentation methods are used as the analyzed methods: k-means clustering, Linear clustering, Adaptive thresholding, Global thresholding. Based on the methods described in the section "Methodology of experiments", the "Global thresholds" method has the best indicators of classification accuracy and time complexity (38.2% - classification accuracy; time complexity is linear (other methods have the same type of complexity, however, GT has much less absolute time indicators).
Keywords: comparison, method, segmentation, image, photo, road, surface, condition, accuracy, classification, time, complextion