Texture Analysis based Detection and Classification of Surface Features on Ageing Infrastructure Elements
Résumé
This paper presents a texture analysis based approach for the detection of damaged regions on the surface of infrastructural elements. A k-means clustering algorithm was used to partition regions with similar textural properties. Four texture measures were derived from a Grey Level Cooccurrence Matrix (GLCM) namely; contrast, homogeneity, entropy and Angular Second Moment (ASM). The approach is validated successfully on an image of a damaged concrete bridge beam. The performance of this Non-Destructive Testing (NDT) technique is evaluated for various values of k through the use of performance points in the Receiver Operating Characteristic (ROC) space. The technique may be deployed as a Structural Health Monitoring (SHM) tool to track the extent of surface damage, and can used in a Bridge Management System (BMS) to aid structured decision making and scheduling of repair work.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...