Aiming at the problem of accuracy degradation caused by illumination change, complex background and fuzzy effect in the extraction of concrete pavement cracks by convolutional neural network, a dynamic index rotation convolution (DIRC) method is proposed. Based on the deformable convolution theory, this method enhances the effectiveness of the index offset by solving the problem that the offset may exceed the receptive field. The dynamic index rotation convolution (DIRC) is introduced into the U2-Net architecture to improve the network's ability to recognize the crack texture of concrete pavement. The results show that on the DeepCrack dataset, compared with the benchmark U2-Net, the F1, Kappa and MIoU indexes of DIRC-U2-Net are increased by 2.40%, 1.30% and 1.49%, respectively. On the CrackForest dataset, the above indicators have increased by 8.43%, 8.47%, and 9.13%, respectively. The visual analysis of the extraction results further shows that the DIRC module significantly enhances the robustness of the U2-Net model to complex interference factors such as illumination differences and image blur. The research conclusions provide a theoretical basis for the accurate and robust extraction of concrete pavement cracks.
LIUYufei, FENGChuqiao, CHENWeile, et al. Review of bridge apparent defect inspection based on machine vision[J]. China Journal of Highway and Transport, 2024, 37(2): 1-15.
[3]
KHERADMANDIN, MEHRANFARV. A critical review and comparative study on image segmentation-based techniques for pavement crack detection[J]. Construction and Building Materials, 2022, 321: 126162.
XUPeng, ZHUXuan, YAODing, et al. Review on intelligent detection and decision-making of asphalt pavement maintenance[J]. Journal of Central South University (Science and Technology), 2021, 52(7): 2099-2117.
[6]
KIRTHIGAR, ELAVENILS. A survey on crack detection in concrete surface using image processing and machine learning[J]. Journal of Building Pathology and Rehabilitation, 2023, 9(1): 15.
[7]
GOLDINGV P, GHARINEIATZ, MUNAWARH S, et al. Crack detection in concrete structures using deep learning[J]. Sustainability, 2022, 14(13): 8117.
HAOMing, LINHuijing, GAOYanyan. Ground fissure extraction method based on improved active contour model for UAV images in mining areas[J]. Journal of Geo-Information Science, 2022, 24(12): 2448-2457.
[10]
QINX B, ZHANGZ C, HUANGC Y, et al. U2-Net: going deeper with nested U-structure for salient object detection[J]. Pattern Recognition, 2020, 106: 107404.
[11]
CHENGH D, LIY J, LIH K, et al. Embankment crack detection in UAV images based on efficient channel attention U2-Net[J]. Structures, 2023, 50: 430-443.
[12]
ZHUX Z, HUH, LINS, et al. Deformable ConvNets V2: more deformable, better results[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 15-20, 2019, Long Beach, CA, USA. IEEE, 2020: 9300-9308.
LIHaifeng, JINGPan, HANHongyang. Airport pavement crack detection algorithm based on deformable convolution and feature fusion[J]. Journal of Nanjing University of Aeronautics & Astronautics, 2021, 53(6): 981-988.
[15]
QIL, LIC H, MEIT. CrackUnet: a novel network with joint network-in-network structure and deformable convolution for pavement crack detection[J]. International Journal of Machine Learning and Cybernetics, 2024, 15(7): 2643-2654.
[16]
WANGC, LIUH B, ANX Y, et al. DCNCrack: pavement crack segmentation based on large-scaled deformable convolutional network[J]. Journal of Computing in Civil Engineering, 2025, 39(2): 04025009.
LIUWeiguang, LIUDong, WANGLu. Survey of deformable convolutional networks[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(7): 1549-1564.
[19]
GUOQ W, WANGC T, XIAOD Q, et al. A novel multi-label pest image classifier using the modified Swin Transformer and soft binary cross entropy loss[J]. Engineering Applications of Artificial Intelligence, 2023, 126: 107060.
[20]
KINGMAD P, BAJ L. Adam: A Method for Stochastic Optimization[C]//BENGIO Y, LECUN Y. 3rd International Conference on Learning Representations, ICLR 2015, Conference Track Proceedings. San Diego, CA, USA: ICLR, 2015.
ZHAOWenhua, LIUAopeng, DUChangbo,et al.Concrete crack detection algorithm based on improved YOLOv8[J].Journal of Liaoning Technical University(Natural Science),2025,44(5):590-596.
[23]
SHIY, CUIL M, QIZ Q, et al. Automatic road crack detection using random structured forests[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 17(12): 3434-3445.
[24]
ONGJ C, LAUS L, ISMADIM Z, et al. Feature pyramid network with self-guided attention refinement module for crack segmentation[J]. Structural Health Monitoring, 2023, 22(1): 672-688.
[25]
ZHANGJ, QIANS R, TANC. Automated bridge surface crack detection and segmentation using computer vision-based deep learning model[J]. Engineering Applications of Artificial Intelligence, 2022, 115: 105225.
[26]
MANJUNATHAP, MASRIS F, NAKANOA, et al. CrackDenseLinkNet: a deep convolutional neural network for semantic segmentation of cracks on concrete surface images[J]. Structural Health Monitoring, 2024, 23(2): 796-817.
[27]
ZHANGY, ZHANGL. Detection of pavement cracks by deep learning models of transformer and UNet[J]. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(11): 15791-15808.
[28]
WANGX F, WANGT K, LIJ L. Advanced crack detection and quantification strategy based on CLAHE enhanced DeepLabv3+[J]. Engineering Applications of Artificial Intelligence, 2023, 126: 106880.