Concrete Crack Classification Using Gray-Level Co-Occurrence Matrix Features Based on Support Vector Machine

Crackas are one of the most common forms of damage occurring in concrete structures, signaling potential structural degradation. Early detection and repair are crucial, as deferred maintenance can lead to economic losses amounting to billions of dollars globally. Structural health monitoring aims to detect and classify defects to enable timely interventions, evaluate protection strategies, and support effective maintenance planning. This study proposed the development of a computer-based system using image processing to assist civil engineers in analyzing concrete surface conditions. The system utilized concrete crack images processed through grayscale conversion and adaptive histogram equalization and applied the Gray Level Co-occurrence Matrix (GLCM) for feature extraction. These features were then classified using the Support Vector Machine (SVM) algorithm. The evaluation included training and testing using 10 -fold cross-validation, and performance was measured with metrics such as accuracy, precision, recall, specificity, and $\mathbf{F}$-score. The results unveiled that the system achieved classification accuracies between 86 % and $\mathbf{9 0 \%}$, demonstrating its potential in automated structural damage assessment.