摘要
为提高隧道衬砌裂缝的识别精度和速度,提出一种基于深度学习的隧道衬砌裂缝自动识别算法。该算法使用两个深度卷积神经网络分别实现隧道衬砌图像分类和裂缝识别。首先使用图像分类网络对隧道衬砌图像进行分类,筛选出含有裂缝的图像;然后使用裂缝识别网络,结合滑动窗口扫描的方式对含有裂缝的图像进行裂缝识别,得到初步的裂缝定位结果;最后根据初步的裂缝定位结果进行图像分割,并计算出裂缝的长度值和宽度值。实验结果表明:图像分类和裂缝识别的准确率均达到98%;处理单张不包含裂缝的大尺度图像耗时0. 008 s,处理单张包含裂缝的图像耗时0. 688 s;施工缝、线缆、字迹等对裂缝识别的影响减弱。
In order to improve the accuracy and speed of tunnel crack recognition,an automatic tunnel lining crack recognition algorithm based on deep learning is proposed. In this algorithm,two deep convolution neural networks are used to classify tunnel lining images and recognize crack damages. Firstly,the tunnel lining images are classified by image classification network,and the images with cracks are screened out. Secondly,the crack identification network is used to identify the cracks in the image with sliding window scanning,and the initial crack location results are obtained. Finally,according to the initial crack location results,the image segmentation is carried out,and the length and width of the crack are calculated. The experimental results show that theaccuracy of image classification and crack recognition is 98%. It takes 0. 008 s to process a single large scale image without cracks,and 0. 688 s to process an image containing a crack. The influence of construction joints,cables and hand writing on the identification of cracks is weakened.
引文
[1] LEE S Y,SANG H L,DONG I S,et al. Development of an inspection system for cracks in a concrete tunnel lining[J].Canadian Journal of Civil Engineering,2007,34(8):966-975.
[2] ZAKERI H,NEJAD F,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):1-43.
[3]朱力强,白彪,王耀东,等.基于特征分析的地铁隧道裂缝识别算法[J].铁道学报,2015,37(5):64-70.
[4]王平让,黄宏伟,薛亚东.基于图像局部网格特征的隧道衬砌裂缝自动识别[J].岩石力学与工程学报,2012,31(5):991-999.
[5] ZHANG L,YANG F,ZHANG Y D,et al. Road crack detection using deep convolutional neural network[C]//IEEE International Conference on Image Processing. Phoenix,USA. IEEE,2016:3708-3712.
[6] CHA Y J,CHOI W,BYKZTRK O. Deep learning-based crack damage detection using convolutional neural networks[J]. Computer-Aided Civil and Infrastructure Engineering,2017,32(5):361-378.
[7] CHEN F C,RMR J. NB-CNN:deep learning-based crack detection using convolutional neural network and Na6ve Bayes data fusion[J]. IEEE Transactions on Industrial Electronics,2018,65(5):4392-4400.
[8] LECUN Y,BOTTOU L,BENGIO Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[9] LECUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature,2015,521(7553):436-44.
[10] KRIZHEVSKY A,SUTSKEVER I,HINTON G. Image Net classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems. San Francisco,USA:Curran Associates Inc,2012:1097-1105.
[11] PANG Y W,SUN M L,JIANG X H,et al. Convolution in convolution for network in network[J]. IEEE Transactions on Neural Networks and Learning Systems,2018,29(5):1587-1597.
[12] MAALOUF A,LARABI M C. Colour image super-resolution using geometric grouplets[J]. Image Processing Iet,2012,6(2):168-180.
[13] REDMON J,DIVVALA S,GIRSHICK R,et al. You only look once:unified,real-time object detection[C]//Computer Vision and Pattern Recognition. Las Vegas,NV,USA:IEEE,2016,1:779-788.
[14] IOFFE S,SZEGEDY C. Batch normalization:accelerating deep network training by reducing internal covariate shift[J].Computer Science,2015(1):448-456.
[15] SHAO Y C,CHEN L C,JAN H H,et al. Measuring system for cracks in concrete using multitemporal images[J]. Journal of Surveying Engineering,2006,132(2):77-82.