Real-Time Crack Detection Using ROV

Abstract

The paper presents a system towards a robust detection model to detect under water concrete cracks on different surfaces and pipelines. The proposed method can preserve different crack’s patterns in different environmental circumstances. Which increases the level of robust detection on concrete and pipelines inspection when combined to Remotely Operated Vehicles (ROV). The system is developed in two phases. Phase one building small size ROV underwater, and phase two developing crack detection model uses low level visual methods based on scale-space decomposition the algorithm is mounted on the ROV to detect the cracks and report the cases to land station. The detection algorithm was tested using a dataset of concrete surfaces classified as positive/negative images with and without cracks were used. The model’s accuracy was verified to identify the best result. For the dataset used in this work, the best experiment yielded a model with accuracy of 92.6%, showcasing the ability of using classical segmentation model for concrete crack detection.