AUTOMATED CRACK EVOLUTION DETECTION FOR STRUCTURAL HEALTH MONITORING

Abstract

Crack detection in civil structures is crucial for ensuring the safety, durability, and functionality of infrastructure such as roads, buildings, and bridges. Traditional methods, including manual inspections, are often time-consuming, costly, and lack accuracy, especially when dealing with large data sets. This dissertation presents advanced algorithms that enhance the efficiency and accuracy of crack detection through innovative image processing techniques.The research introduces a novel clustering algorithm that eliminates the need for arbitrary threshold selection and minimizes manual intervention by intelligently grouping pixels in crack images. This method offers greater versatility and adaptability compared to traditional approaches, effectively detecting crack sizes with high precision. Building upon this, the study integrates advanced feature matching techniques to address challenges faced by conventional image-based methods. This approach enhances efficiency and significantly reduces noise, blemishes, and other image disturbances, while accurately detecting crack growth regardless of image scale and orientation. Through labeling, dilation, and superposition processes, the algorithm eliminates noise and ensures small and thin cracks are not overlooked. Furthermore, the research incorporates robust estimation algorithms for homography estimation to correct perspective distortion issues commonly encountered in images captured by unmanned aerial vehicles (UAVs). This innovative approach substantially improves the precision and reliability of crack detection by autonomously utilizing natural image features for perspective correction, effectively navigating geometric complexities. Comprehensive analyses and comparisons demonstrate that the proposed methods significantly outperform existing crack detection techniques in terms of accuracy and efficiency. These approaches contribute to the advancement of structural health monitoring (SHM) practices, enhancing both theoretical and practical aspects of maintaining and ensuring the safety of civil infrastructures worldwide.