A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Concrete Surface Defect Image Analysis Based on Novel Deep Learning Algorithm
During the paving and commissioning of concrete pavements, it is often found that defects of the same type show great differences in appearance and location on the different concretes, while defects of different types show great similarities on concretes. Therefore, how to find the commonalities and accurately detect defects is the focus of this study. Under this condition, this study proposes the novel concrete surface defect image analysis framework based on the novel deep learning algorithm. To begin with, the proposed framework includes an image crack detection module, which effectively combines the enhanced OTSU algorithm and gradient-based edge detection technology to improve the detection effect of crack areas in noisy and degraded scenes to a certain extent, thereby obtaining stable crack detection results. Then, the framework also proposes a concrete surface defect image enhancement model, which corrects the influence of uneven lighting by reviewing the illumination reflection model and Gaussian function convolution. This action is aiming to meet the detection needs of different scenes. Finally, YOLOv5 model enhanced by EDNet+ is proposed for the final surface defect image analysis. The simulation and experiment reflect that, compared with 2 state-of-the-art models, the proposed framework performs better.
Concrete Surface Defect Image Analysis Based on Novel Deep Learning Algorithm
During the paving and commissioning of concrete pavements, it is often found that defects of the same type show great differences in appearance and location on the different concretes, while defects of different types show great similarities on concretes. Therefore, how to find the commonalities and accurately detect defects is the focus of this study. Under this condition, this study proposes the novel concrete surface defect image analysis framework based on the novel deep learning algorithm. To begin with, the proposed framework includes an image crack detection module, which effectively combines the enhanced OTSU algorithm and gradient-based edge detection technology to improve the detection effect of crack areas in noisy and degraded scenes to a certain extent, thereby obtaining stable crack detection results. Then, the framework also proposes a concrete surface defect image enhancement model, which corrects the influence of uneven lighting by reviewing the illumination reflection model and Gaussian function convolution. This action is aiming to meet the detection needs of different scenes. Finally, YOLOv5 model enhanced by EDNet+ is proposed for the final surface defect image analysis. The simulation and experiment reflect that, compared with 2 state-of-the-art models, the proposed framework performs better.
Concrete Surface Defect Image Analysis Based on Novel Deep Learning Algorithm
Dong, Naibao (author) / Li, Xinkai (author) / Ren, Jingfei (author) / Xin, Xin (author)
2025-01-20
946640 byte
Conference paper
Electronic Resource
English
Deep Learning-Based Intelligent Detection Algorithm for Surface Disease in Concrete Buildings
| DOAJ | 2024
Rail Surface Defect Detection Method Based on Deep Learning Method with 3D Range Image
| Springer Verlag | 2023