Accurate detection of bridge cracks is crucial for ensuring traffic safety, but the slender morphology of cracks and complex backgrounds bring great challenges to high-precision detection. A bridge crack segmentation network, WMCNet, was proposed, which fused wavelet transform and the Mamba architecture to enhance crack detection performance in complex scenarios by coordinating frequency-domain information and spatial context. A wavelet Mamba interaction module was adopted to decompose features into multiple frequency bands, and the Mamba structure was utilized to enhance its long-range dependencies to capture global information. An edge-enhancement feature fusion module was designed, which performed deep fusion and boundary optimization on multi-scale features in skip connections by introducing a fuzzy graph inference mechanism that utilized graph networks and fuzzy logic. To further optimize the segmentation results, the network training adopted a hybrid loss function that combines BCE and Dice losses to ensure pixel-level accuracy, and an adaptive loss module was introduced to enhance the model's feature representation capability. Experiments demonstrate that WMCNet achieves mIoU scores of 88.58%, 86.53%, and 83.34% on the self-built CQBCD dataset and the public crack datasets DeepCrack and CrackTree260, respectively. Compared with nine other models, the proposed network exhibits superior performance.
LAN Zhang-li, HE Chuan-han, ZHANG Hong, HUANG Wei-hong, MA Xin, DU Song, DENG Guo-bing.
Bridge Crack Segmentation Network Based on Wavelet-Mamba Collaboration and Edge Enhancement[J].
Science Technology and Engineering, 2026, 26(22): 9618-9628 DOI:10.12404/j.issn.1671-1815.2505964
状态空间模型(state space model,SSM)近年来在长序列建模领域展现出突破性优势,其线性计算复杂度与全局依赖建模能力的独特结合,为视觉任务提供了超越传统CNN和Transformer的新范式[17]。Mamba通过选择性状态空间机制(S6)实现了输入自适应的动态特征选择,在保持线性复杂度的同时,显著提升了远程上下文建模能力[18]。
SongZegang, LiuYanli, ZhangChangxing. Review on the application and development of bridge crack detection based on machine vision[J]. Science Technology and Engineering, 2023, 23(30): 12796-12805.
LiuYuanyuan, ChenJiahao, WangJingzhi, et al. Lightweight pavement crack segmentation network based on multi-branch feature extraction[J]. Journal of Traffic and Transportation Engineering, 2025, 25(3): 205-220.
ZhuJunbin, DuBin, XuShimin. Crack detection based on deep supervision network fusing multi-scale and multi-channel features[J]. Science Technology and Engineering, 2024, 24(13): 5595-5603.
LiDongfeng, ChenYuren, YuBo. Pavement crack detection met-hod based on multi-level feature fusion[J]. Computer Engineering, 2026, 52(1): 154-165.
[11]
YangL, HuangH, KongS, et al. A deep segmentation network for crack detection with progressive and hierarchical context fusion[J]. Journal of Building Engineering, 2023, 75: 106886.
ChuHonghu, YuanHuaqing, LongLizhi, et al. A Transformer based cascade method for segmenting bridge cracks from high-resolution images[J]. China Journal of Highway and Transport, 2024, 37(2): 65-76.
[14]
GuoF, LiuJ, LüC, et al. A novel transformer-based network with attention mechanism for automatic pavement crack detection[J]. Construction and Building Materials, 2023, 391: 131852.
[15]
XuG, LiaoW, ZhangX, et al. Haar wavelet downsampling: a simple but effective downsampling module for semantic segmentation[J]. Pattern Recognition, 2023, 143: 109819.
[16]
SunR, LiX, ZhangL, et al. Wavelet-integrated deep neuralnetwork for deblurring and segmentation of crack images[J]. Mechanical Systems and Signal Processing, 2025, 224: 112240.
ChengRong, ZhuWenzhong, WangWen. UltraLight CrackNet: lightweight crack segmentation network based on VMamba[J]. Electronic Measurement Technology, 2025(22): 224-234.
[19]
TangW, WuZ, WangW, et al. VM-Unet++ research on crack image segmentation based on improved VM-Unet[J]. Scientific Reports, 2025, 15(1): 8938.
[20]
FinderS E, AmoyalR, TreisterE, et al. Wavelet convolutions for large receptive fields[C]// European Conference on Computer Vision. Cham: Springer Nature, 2024: 363-380.
LiuYunqing, WuYue, ZhangQiong, et al. Road crack detection based on the fusion of separable convolution and wavelet transform[J]. Computer Science, 2024, 51(S2): 314-322.
[23]
GengP, LuJ, MaH, et al. Crack segmentation based on fusing multi-scale wavelet and spatial-channel attention[J]. Structural Durability & Health Monitoring, 2023, 17(1): 1-22.
[24]
GuoL, XiongF, CaoY, et al. Focusing on cracks with instance normalization wavelet layer[J]. Sensors, 2024, 25(1): 146.
[25]
ZhangH, ZhuY, WangD, et al. A survey on visual mamba[J]. Applied Sciences, 2024, 14(13): 5683.
[26]
GuA, DaoT. Mamba: linear-time sequence modeling with selective state spaces[J]. arXiv Preprint arXiv: 2312.00752, 2023.
[27]
ZuoX, ShengY, ShenJ, et al. Topology-aware Mamba for crack segmentation in structures[J]. Automation in Construction, 2024, 168: 105845.
[28]
HanC, YangH, YangY. Enhancing pixel-level crack segmentation with visual mamba and convolutional networks[J]. Automation in Construction, 2024, 168: 105770.
[29]
ZhangZ, PengB, ZhaoT. An ultra-lightweight network combining Mamba and frequency-domain feature extraction for pavement tiny-crack segmentation[J]. Expert Systems with Applications, 2025, 264: 125941.
LiuH, JiaC, ShiF, et al. SCSegamba: lightweight structure-aware vision mamba for crack segmentation in structures[C]// Proceedings of the Computer Vision and Pattern Recognition Conference. Nashville: CVPR, 2025: 29406-29416.
[32]
YangL, WangS, LiewA W C. Fine-grained lip image segmentation using fuzzy logic and graph reasoning[J]. IEEE Transactions on Fuzzy Systems, 2023, 32(2): 349-359.
[33]
LiuY, YaoJ, LuX, et al. DeepCrack: a deep hierarchical feature learning architecture for crack segmentation[J]. Neurocomputing, 2019, 338: 139-153.
[34]
WangF, JiJ, WangY. Remote sensing image semantic segmentation based on cascaded Transformer[J]. IEEE Transactions on Artificial Intelligence, 2024, 5(8): 4136-4148.
[35]
ZouQ, ZhangZ, LiQ, et al. Deepcrack: learning hierarchical convolutional features for crack detection[J]. IEEE Transactions on Image Processing, 2018, 28(3): 1498-1512.
[36]
RonnebergerO, FischerP, BroxT. U-net: convolutional networks for biomedical image segmentation[C]// Medical Image Computing and Computer-assisted Intervention-MICCAI 2015: 18th International Conference. Berlin: Springer, 2015: 234-241.
[37]
ChenJ, LuY, YuQ, et al. Transunet: Transformers make strong encoders for medical image segmentation[J]. arXiv Preprint arXiv: 2102.04306, 2021.
[38]
ZhaoH, ShiJ, QiX, et al. Pyramid scene parsing network[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2017: 2881-2890.
[39]
BadrinarayananV, KendallA, CipollaR. Segnet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495.
[40]
ChenL C, ZhuY, PapandreouG, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]// Proceedings of the European Conference on Computer Vision (ECCV). Munich: ECCV, 2018: 801-818.
[41]
TanM, LeQ. Efficientnet: rethinking model scaling for convolutional neural networks[C]// International Conference on Machine Learning. Long Beach: PMLR, 2019: 6105-6114.
[42]
LiuJ, YangH, ZhouH Y, et al. Swin-umamba: Mamba-based unet with imagenet-based pretraining[C]// International Confere-nce on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2024: 615-625.
[43]
PangJ, ZhangH, ZhaoH, et al. DcsNet: a real-time deep network for crack segmentation[J]. Signal, Image and Video Processing, 2022, 16(4): 1-9.