混凝土作为公路建设的重要材料，由于长期受外部环境侵蚀，易出现道路裂缝等病害，从而导致严重的交通事故，存在极大的安全隐患，因此对混凝土路面裂缝进行检测和量化评估对道路修补具有重要的现实意义。现阶段深度学习方法已经成为完成混凝土路面裂缝相关任务的主要方式，通过充分学习路面裂缝特征，能较好地适应混凝土路面图像中复杂的背景，并且基于深度学习的混凝土路面裂缝检测算法拥有较高的准确率。

通过分析混凝土路面裂缝图像中道路背景与裂缝之间的特征差异，本文提出了一种基于DeepLabV3Plus的路面裂缝像素级语义分割算法。为提高对路面裂缝的提取速度和分割效果，将主干网络Modified Aligned Xception替换为轻量级网络MobileNetV2，加速提取裂缝图像特征；同时将ASPP模块由并行连接调整为级联拼接，扩大模型感受野，并增强各支路的相关性；此外，在解码器结构中融合了两个浅层特征，为分割结果提供了更多细节信息。针对图像中道路背景与裂缝存在类不平衡的问题，算法采用相应类别加权的方式对损失函数进行修正。改进的模型在本文构建的路面裂缝数据集上MIoU和MAP分别达到了75.18%和83.64%，单张图片的分割时间缩短为58.2ms，与原本分割模型相比，分割速度大幅提升，表明该模型能够快速有效地对路面裂缝进行分割提取。

为了在路面裂缝提取结果的基础上对其进行分类与量化评估，本文提出了基于MobileNetV2的路面裂缝分类模型，同时建立路面裂缝评价体系。分类模型融合了多尺度特征和CBAM注意力机制，加强对裂缝图像有效信息的关注，提升裂缝图像分类的精度。在路面评价体系中，分别对裂缝长度、平均宽度、面积以及破损比率进行计算，从而量化裂缝参数，并根据路面等级划分标准对路面状态进行评估。实验结果表明，本文分类模型在裂缝二值图像分类数据集的准确率达到了90.81%，相比MobileNetV2提高了2.79%，模型能够快速的对裂缝二值图像进行分类。此外，根据对路面裂缝的量化结果和相对误差率，表明本文对裂缝的提取较为准确，符合混凝土道路裂缝的实际情况。

Concrete is an important material for highway construction. Due to the long-term erosion of the external environment, road cracks and other diseases are prone to occur, resulting in serious traffic accidents and great potential safety hazards. Therefore, the detection and quantitative evaluation of concrete pavement cracks is of great significance for road repair. important practical significance. At present, the deep learning method has become the main way to complete the tasks related to concrete pavement cracks. By fully learning the characteristics of pavement cracks, it can better adapt to the complex background in the concrete pavement image, and the concrete pavement crack detection algorithm based on deep learning has high 's accuracy.

By analyzing the feature differences between road background and cracks in concrete pavement crack images, this paper proposes a pixel-level semantic segmentation algorithm for pavement cracks based on DeepLabV3Plus. In order to improve the extraction speed and segmentation effect of pavement cracks, the backbone network Modified Aligned Xception is replaced with a lightweight network MobileNetV2 to accelerate the extraction of crack image features; At the same time, the ASPP module is adjusted from parallel connection to cascade splicing, which expands the receptive field of the model and enhances the correlation of each branch; Furthermore, two shallow features are fused in the decoder structure, which provides more detailed information for the segmentation results. Aiming at the problem of class imbalance between the road background and cracks in the image, the algorithm uses the corresponding class weighting method to correct the loss function. The improved model achieves MIoU and MAP of 75.18% and 83.64% respectively on the pavement crack dataset constructed in this paper, and the segmentation time of a single image is shortened to 58.2ms. Compared with the original segmentation model, the segmentation speed is greatly improved, indicating that the model It can quickly and effectively segment and extract pavement cracks.

In order to classify and quantitatively evaluate pavement cracks based on the extraction results, this paper proposes a pavement crack classification model based on MobileNetV2, and establishes a pavement crack evaluation system. The classification model integrates multi-scale features and CBAM attention mechanism, strengthens the attention to the effective information of crack images, and improves the accuracy of crack image classification. In the pavement evaluation system, the crack length, average width, area and damage ratio are calculated respectively to quantify the crack parameters, and the pavement condition is evaluated according to the pavement grade classification standard. The experimental results show that the accuracy of the classification model in this paper in the crack binary image classification dataset reaches 90.81%, which is 2.79% higher than that of MobileNetV2. The model can quickly classify crack binary images. In addition, according to the quantitative results of pavement cracks and the relative error rate, it shows that the extraction of cracks in this paper is more accurate, which is in line with the actual situation of concrete pavement cracks.

[1]刘颖, 关昌余. 全力建立现代化公路养护管理体系——解读《“十三五” 公路养护管理发展纲要》[J]. 中国公路, 2016 (15): 54-58.

[2]Mohan A, Poobal S. Crack detection using image processing: A critical review and analysis[J]. Alexandria Engineering Journal, 2018, 57(2): 787-798.

[3]Zakeri H, Nejad F M, Fahimifar A. Image based techniques for crack detection, classification and quantification in asphalt pavement: a review[J]. Archives of Computational Methods in Engineering, 2017, 24(4): 935-977.

[4]Alshemali B, Kalita J. Improving the reliability of deep neural networks in NLP: A review[J]. Knowledge-Based Systems, 2020, 191: 105210.

[5]Nassif A B, Shahin I, Attili I, et al. Speech recognition using deep neural networks: A systematic review[J]. IEEE access, 2019, 7: 19143-19165.

[6]Deng J, Guo J, Xue N, et al. Arcface: Additive angular margin loss for deep face recognition[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4690-4699.

[7]Scherhag U, Rathgeb C, Merkle J, et al. Face recognition systems under morphing attacks: A survey[J]. IEEE Access, 2019, 7: 23012-23026.

[8]Chauhan R, Ghanshala K K, Joshi R C. Convolutional neural network (CNN) for image detection and recognition[C]//2018 First International Conference on Secure Cyber Computing and Communication (ICSCCC). IEEE, 2018: 278-282.

[9]Minaee S, Boykov Y Y, Porikli F, et al. Image segmentation using deep learning: A survey[J]. IEEE transactions on pattern analysis and machine intelligence, 2021.

[10]Wang G, Peter W T, Yuan M. Automatic internal crack detection from a sequence of infrared images with a triple-threshold Canny edge detector[J]. Measurement Science and Technology, 2018, 29(2): 025403.

[11]Ahmed A S. Comparative study among Sobel, Prewitt and Canny edge detection operators used in image processing[J]. J. Theor. Appl. Inf. Technol, 2018, 96(19): 6517-6525.

[12]周慧媛, 邱书波, 刘海英, 等. 基于对比度受限自适应直方图多种路面裂缝检测与识别[J].2018.

[13]彭刚, 刘博. 变电站智能巡检系统墙体裂缝识别[J]. 智能机器人, 2017 (4): 53-55.

[14]Peng L, Chao W, Shuangmiao L, et al. Research on crack detection method of airport runway based on twice-threshold segmentation[C]//2015 Fifth International Conference on Instrumentation and Measurement, Computer, Communication and Control (IMCCC). IEEE, 2015: 1716-1720.

[15]李海丰, 吴治龙, 聂晶晶. 强干扰条件下机场道面细小裂缝自动识别算法[J]. 计算机工程与科学, 2020.

[16]Medina R, Llamas J, Gómez-García-Bermejo J, et al. Crack detection in concrete tunnels using a gabor filter invariant to rotation[J]. Sensors, 2017, 17(7): 1670.

[17]Ai D, Jiang G, Kei L S, et al. Automatic pixel-level pavement crack detection using information of multi-scale neighborhoods[J]. IEEE Access, 2018, 6: 24452-24463.

[18]李清泉, 邹勤, 张德津. 利用高精度三维测量技术进行路面破损检测[J]. 武汉大学学报: 信息科学版, 2017, 42(11): 1549-1564.

[19]韩锟, 韩洪飞. 基于区域级和像素级特征的路面裂缝检测方法[J]. 铁道科学与工程学报, 2018, 15(5): 1178-1186.

[20]马芸, 王国军. 基于二维复数离散小波包变换的桥面裂缝检测[J]. 沈阳工业大学学报, 2018, 40(6): 659-663.

[21]Akagic A, Buza E, Omanovic S, et al. Pavement crack detection using Otsu thresholding for image segmentation[C]//2018 41st International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO). IEEE, 2018: 1092-1097.

[22]Luo Q, Ge B, Tian Q. A fast adaptive crack detection algorithm based on a double-edge extraction operator of FSM[J]. Construction and Building Materials, 2019, 204: 244-254.

[23]Chen F C, Jahanshahi M R. NB-CNN: Deep learning-based crack detection using convolutional neural network and Naïve Bayes data fusion[J]. IEEE Transactions on Industrial Electronics, 2017, 65(5): 4392-4400.

[24]Hoang N D, Nguyen Q L, Tien Bui D. Image processing–based classification of asphalt pavement cracks using support vector machine optimized by artificial bee colony[J]. Journal of Computing in Civil Engineering, 2018, 32(5): 04018037.

[25]Liu Y, Yao J, Lu X, et al. DeepCrack: A deep hierarchical feature learning architecture for crack segmentation[J]. Neurocomputing, 2019, 338: 139-153.

[26]曹锦纲, 杨国田, 杨锡运. 基于注意力机制的深度学习路面裂缝检测[J]. 计算机辅助设计与图形学学报, 2020, 32(8): 1324-1333.

[27]Chen T, Cai Z, Zhao X, et al. Pavement crack detection and recognition using the architecture of segNet[J]. Journal of Industrial Information Integration, 2020, 18: 100144.

[28]Li S, Zhao X, Zhou G. Automatic pixel‐level multiple damage detection of concrete structure using fully convolutional network[J]. Computer‐Aided Civil and Infrastructure Engineering, 2019, 34(7): 616-634.

[29]Zhang K, Zhang Y, Cheng H D. Crackgan: Pavement crack detection using partially accurate ground truths based on generative adversarial learning[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 22(2): 1306-1319.

[30]Yang X, Li H, Yu Y, et al. Automatic pixel‐level crack detection and measurement using fully convolutional network[J]. Computer‐Aided Civil and Infrastructure Engineering, 2018, 33(12): 1090-1109.

[31]Fang J, Qu B, Yuan Y. Distribution equalization learning mechanism for road crack detection[J]. Neurocomputing, 2021, 424: 193-204.

[32]Islam M, Sohaib M, Kim J, et al. Crack classification of a pressure vessel using feature selection and deep learning methods[J]. Sensors, 2018, 18(12): 4379.

[33]Fang F, Li L, Rice M, et al. Towards real-time crack detection using a deep neural network with a Bayesian fusion algorithm[C]//2019 IEEE International Conference on Image Processing (ICIP). IEEE, 2019: 2976-2980.

[34]Bibi R, Saeed Y, Zeb A, et al. Edge AI-based automated detection and classification of road anomalies in VANET using deep learning[J]. Computational intelligence and neuroscience, 2021, 2021.

[35]Azhari F, Sennersten C, Milford M, et al. PointCrack3D: Crack Detection in Unstructured Environments using a 3D-Point-Cloud-Based Deep Neural Network[J]. arXiv preprint arXiv:2111.11615, 2021.

[36]UKAI M. Tunnel lining crack detection method by means of deep learning[J]. Quarterly Report of RTRI, 2019, 60(1): 33-39.

[37]Guo L, Li R, Jiang B, et al. Automatic crack distress classification from concrete surface images using a novel deep-width network architecture[J]. Neurocomputing, 2020, 397: 383-392.

[38]Guo L, Li R, Jiang B. A cascade broad neural network for concrete structural crack damage automated classification[J]. IEEE Transactions on Industrial Informatics, 2020, 17(4): 2737-2742.

[39]Tang Y, Zhang A A, Luo L, et al. Pixel-level pavement crack segmentation with encoder-decoder network[J]. Measurement, 2021, 184: 109914.

[40]Nguyen N H T, Perry S, Bone D, et al. Two-stage convolutional neural network for road crack detection and segmentation[J]. Expert Systems with Applications, 2021, 186: 115718.

[41]Yu F, Koltun V. Multi-scale context aggregation by dilated convolutions[J]. arXiv preprint arXiv:1511.07122, 2015.

[42]Sandler M, Howard A, Zhu M, et al. Mobilenetv2: Inverted residuals and linear bottlenecks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 4510-4520.

[43]Howard A G, Zhu M, Chen B, et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications[J]. arXiv preprint arXiv:1704.04861, 2017.

[44]He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[45]Chen L C, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 801-818.

[46]Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[J]. arXiv preprint arXiv:1706.05587, 2017.

[47]Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]//International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015: 234-241.

[48]Chollet F. Xception: Deep learning with depthwise separable convolutions[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 1251-1258.

[49]Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 2818-2826.

[50]JTG 5142—2019. 公路沥青路面养护技术规范[S]. 北京: 人民交通出版社, 2019.

[51]JTG H20-2018. 公路技术状况评定标准[S]. 北京: 人民交通出版社, 2018.

[52]马莹双. 基于轻量级CNN的垃圾分类识别算法研究[D]. 西安科技大学, 2021.

[53]孟俊熙, 张莉, 曹洋, 张乐天, 宋倩. 基于Deeplab v3+的图像语义分割算法优化研究[J/OL]. 激光与光电子学进展: 1-15[2022-03-16].

[54]郭梦利, 阮顺领, 卢才武, 顾清华.基于改进DeepLabv3+网络的露天矿路网提取方法[J]. 激光与光电子学进展, 2021, 58(22): 477-486.

[55]Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2881-2890.

[56]Badrinarayanan V, Kendall A, SegNet R C. A deep convolutional encoder-decoder architecture for image segmentation[J]. arXiv preprint arXiv:1511.00561, 2015, 5.

[57]Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 7132-7141.

[58]Li X, Hu X, Yang J. Spatial group-wise enhance: Improving semantic feature learning in convolutional networks[J]. arXiv preprint arXiv:1905.09646, 2019.

[59]Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.

[60]贾国辉. 基于多尺度卷积特征的路面裂缝检测方法研究[D]. 辽宁工程技术大学, 2020.

[61]孙朝云, 马志丹, 李伟,等. 基于深度卷积神经网络融合模型的路面裂缝识别方法[J]. 长安大学学报: 自然科学版, 2020, 40(4): 1-13.

[62]Ji A, Xue X, Wang Y, et al. An integrated approach to automatic pixel-level crack detection and quantification of asphalt pavement[J]. Automation in Construction, 2020, 114: 103176.

[63]Qu Z, Mei J, Liu L, et al. Crack detection of concrete pavement with cross-entropy loss function and improved VGG16 network model[J]. Ieee Access, 2020, 8: 54564-54573.