Comparison of image edge detection methods on potholes road images. (August 2020)

• Record Type: Journal Article
• Authors: Indriyani, Tutuk; Utoyo, Imam; Rulaningtyas, Riries
• Is Part Of: Journal of physics. Volume 1613(2020)
• Publication Date: 2020-08
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1613/1/012067
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Dinas pekerjaan Umum (DPU) in Surabaya in carrying out road repairs, especially in potholes, must know the position, area and depth. So it is important to do research to find out the surface area of potholes. To find out the surface area of the hole, the edge detection process is done first. Edge detection functions to get the edge of an object. Edge detection is obtained by utilizing a drastic change in the intensity value at the boundary of two areas. In this study, to determine the edge of an object (potholes road) by comparing three edge detection methods consisting of the Frei-Chen, Laplacian and Laplacian of Gaussian (LoG) methods. From the results of testing the Laplacian of Gaussian method has an average value of accuracy of 67%, sensitivity of 81.97% and specificity of 65.16%.The measurement results obtained by the Lapalacian of Gaussian method which is the best method for edge detection in potholes road images, because it produces the highest accuracy, sensitivity and specificity.