A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
A high precision crack classification system using multi-layered image processing and deep belief learning
Road surfaces experience fatigue stress and loading, which often lead to cracks on the surface. The cracks might cause serious damage, and therefore, early detection can reduce the road maintenance cost. Traditional inspection methods are carried out by humans and are slow, costly and hazardous. To improve accuracy and reduce the hazards of current crack detection methods, this paper proposes a new autonomous crack detection system (ACDS) that can be used in any autonomous vehicles (UAVs). ACDS consists of three stages: image acquisition, image processing, and classification. The image processing stage consists of five parallel filtering methods, which remove noise and extract features from the images. In the classification stage, five deep belief network (DBN) classifiers separately analyse the images to detect cracks. The dataset used in this paper contains 15,000 RGB and infrared images, with or without cracks. The results show the high precision of the proposed system.
A high precision crack classification system using multi-layered image processing and deep belief learning
Road surfaces experience fatigue stress and loading, which often lead to cracks on the surface. The cracks might cause serious damage, and therefore, early detection can reduce the road maintenance cost. Traditional inspection methods are carried out by humans and are slow, costly and hazardous. To improve accuracy and reduce the hazards of current crack detection methods, this paper proposes a new autonomous crack detection system (ACDS) that can be used in any autonomous vehicles (UAVs). ACDS consists of three stages: image acquisition, image processing, and classification. The image processing stage consists of five parallel filtering methods, which remove noise and extract features from the images. In the classification stage, five deep belief network (DBN) classifiers separately analyse the images to detect cracks. The dataset used in this paper contains 15,000 RGB and infrared images, with or without cracks. The results show the high precision of the proposed system.
A high precision crack classification system using multi-layered image processing and deep belief learning
Jo, Jun (author) / Jadidi, Zahra (author)
Structure and Infrastructure Engineering ; 16 ; 297-305
2020-02-01
9 pages
Article (Journal)
Electronic Resource
English
Learning to Diversify Deep Belief Networks for Hyperspectral Image Classification
| Online Contents | 2017
Crack detection and classification in asphalt pavement using image processing
| British Library Conference Proceedings | 2008
Vehicle Detection and Classification via YOLOv8 and Deep Belief Network over Aerial Image Sequences
| DOAJ | 2023