Addressing the challenges posed by the complexity of UAV image scenes, significant differences in view angles, and frequent occlusions by foreign objects, an automatic extraction method of UAV multi-view image rail lines based on CBAS_Unet is proposed. On the basis of the traditional U-net network, a parallel Atrous Spatial Pyramid Pooling (ASPP) and Convolutional Block Attention Module (CBAM) are added to enhance the network's capability to capture contextual information across various scales, thereby significantly boosting the segmentation performance of rails. High-precision extraction of complete rail vector lines is then achieved by pixel grouping using RANSAC least squares fitting and chaining of neighboring rail lines. The experimental results show that, compared with the two classical models of U-net and Deeplab v3+, the proposed method achieves an increase in the intersection and merger ratio for rail segmentation of multi-view UAV images by 2.09% and 1.98%, and the score value is improved by 1.50 and 1.42, respectively. The completeness of rail line extraction reaches 90.7%, surpassing the 83.3% achieved by the U-net model. The average error in rail line extraction is approximately 0.58 pixels, with the median error of approximately 0.77 pixels, enabling sub-pixel-level rail line extraction. This method fulfills the requirements for automation, comprehensive and high-precision extraction of rail lines from UAV multi-view images.
WANGGuangshuai. High-Precision Automatic Reconstruction of Railway Line with UAV Oblique Photography [J]. Bulletin of Surveying and Mapping, 2022 (5): 133-139, 156. in Chinese
DENGJiwei. Measurement Method for the Turnout Center of Existing Railway Line Based on High-Precision True Orthophoto [J]. China Railway Science, 2023, 44 (2): 24-31. in Chinese
[5]
SHELHAMERE, LONGJ, DARRELLT. Fully Convolutional Networks for Semantic Segmentation [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (4): 640-651.
[6]
ZHANGS C, WANGC Y, LIJ H, et al. MF-Dfnet: a Deep Learning Method for Pixel-Wise Classification of Very High-Resolution Remote Sensing Images [J]. International Journal of Remote Sensing, 2022, 43 (1): 330-348.
[7]
CHENH, LIUY L, WANGD C, et al. Automatic Extraction of Bare Soil Land from High-Resolution Remote Sensing Images Based on Semantic Segmentation with Deep Learning [J]. Remote Sensing, 2023, 15 (6): 1646.
[8]
CHENZ Y, LID L, FANW T, et al. Self-Attention in Reconstruction Bias U-Net for Semantic Segmentation of Building Rooftops in Optical Remote Sensing Images [J]. Remote Sensing, 2021, 13 (13): 2524.
LUTong, YUZujun, GUOBaoqing, et al. Semantic Segmentation of Railway Scene Based on Reticulated Multi-Scale and Bidirectional Channel Attention [J]. Journal of Transportation Systems Engineering and Information Technology, 2023, 23 (2): 233-241, 299. in Chinese
WANGHui, WUYujie, FANZizhu, et al. Fast Detection Algorithm of Railway Clearance Based on Deep Learning [J]. Journal of Railway Science and Engineering, 2023, 20 (4): 1223-1231. in Chinese
[13]
LANM, ZHANGY P, ZHANGL F, et al. Global Context Based Automatic Road Segmentation via Dilated Convolutional Neural Network [J]. Information Sciences, 2020, 535: 156-171.
SHENYu, WANGHailong, LIANGDong, et al. River Extraction Method from Remote Sensing Images of Cold and Arid Regions Based on Self-Supervised Comparative Learning [J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54 (6): 125-135. in Chinese
ZHANGWeiguang, ZHONGJingtao, HUYanju, et al. Extraction and Quantification of Pavement Alligator Crack Morphology Based on VGG16-UNet Semantic Segmentation Model [J]. Journal of Traffic and Transportation Engineering, 2023, 23 (2): 166-182. in Chinese
ZHANGZhihua, WENYanan, MUHaowei, et al. Dual Attention Mechanism Based Pavement Crack Detection [J]. Journal of Image and Graphics, 2022, 27 (7): 2240-2250. in Chinese
WANGWeidong, ZHANGChenlei, HUWenbo, et al. Fine-Grained Measurement of Ballastless Track Slab Cracks Based on Improved Faster R-CNN and Orthogonal Projection [J]. China Railway Science, 2023, 44 (6): 46-56. in Chinese
[22]
RONNEBERGERO, FISCHERP, BROXT. U-Net: Convolutional Networks for Biomedical Image Segmentation [C]// Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015. Lecture Notes in Computer Science, 9351. Cham: Springer, 2015: 234-241.
[23]
CHENL C, ZHUY K, PAPANDREOUG, et al. Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation [C]// Computer Vision ECCV 2018. Lecture Notes in Computer Science, 11211. Cham, 2018: 833-851.
[24]
CHENB, ZHANGZ, LIUN, et al. Spatiotemporal Convolutional Neural Network with Convolutional Block Attention Module for Micro-Expression Recognition [J]. Information, 2020, 11 (8):380.
[25]
LIX Y, SUNX F, MENGY X, et al. Dice Loss for Data-Imbalanced NLP Tasks [EB/OL]. 2019: arXiv: 1911.02855.
[26]
LINT Y, GOYALP, GIRSHICKR, et al. Focal Loss for Dense Object Detection [C]// 2017 IEEE International Conference on Computer Vision. Venice, New York: IEEE Press, 2017: 2999-3007.