A framework for crack detection based on sensor integration 3D model for the maintenance of old structures

Abstract

The existing human-centered safety diagnosis method for structure maintenance has problems with time, cost, accuracy, and safety. Recently, many attempts have been made to quickly and accurately check the condition of structures for maintenance by utilizing equipment. In this research, a framework was established to derive the priority of repairing cracks based on a 3D model after data was obtained by combining LiDAR, high-resolution camera, and thermal imaging camera with drones. Drones serve as a moving object for obtaining data and can be used in places where it is difficult to access with manpower, thereby improving the safety of workers. The framework is as follows: Step 1 is obtaining data of a structure through each sensor, which is classified according to two purposes. In Step 1-1, Point Cloud Data is obtained through LiDAR, and crack images of the structure are obtained through high-resolutions camera. In this step, data can be obtained to determine the accurate location and size of cracks in the structure. In Step 1-2, thermal imaging data for measuring the depth of cracks is obtained through thermal imaging camera. Step 2 is processing and integrating data obtained through each sensor for 3D visualization of cracks on the structure. The PCD obtained through LiDAR is made into a mesh after removing noise, and a 3D model is constructed through texturing using high-resolution images and thermal images on the mesh, respectively. Step 3 is deriving the repair priority of cracks identified on the 3D model. Each crack has different characteristics such as location, size, and depth. For crack repair, cracks are classified according to grade and the repair priority is determined by considering the characteristics of crack. The framework can contribute to improving the economic efficiency, accuracy, and safety of structure maintenance by presenting criteria for deriving crack repair priorities.