House extraction from railway remote sensing images based on improved Mask R-CNN model
-
摘要: 针对目前铁路建设预可行性研究阶段地形图制作存在的人工目视遥感解译效率低、生产周期长等问题，结合深度学习的特点和优势，对实例分割模型Mask R-CNN（Mask Region-based Convolutional Neural Network）进行改进，选用ResNeXt101作为其主干特征提取网络，并利用边缘提取算法进一步实现了遥感影像的自动矢量化提取。试验结果表明，改进后的模型在Mask AP50、Box AP50、Mask mAP、Box mAP等指标上均有明显的提升，可生成供铁路建设预可行性研究阶段拆迁费用计算的房屋矢量化影像，为该阶段的影像处理工作提供技术支撑。
-
关键词:
- 铁道工程 /
- 模型改进 /
- 实例分割 /
- Mask R-CNN /
- 遥感影像
Abstract: In response to the low efficiency of manual visual remote sensing interpretation and long production cycle in the production of topographic maps during the pre-feasibility study stage of railway construction, this paper combined the characteristics and advantages of deep learning to improve the instance segmentation model Mask R-CNN (Mask Region based Convolutional Neural Network). The paper selected ResNeXt101 as its backbone feature extraction network, and further implemented automatic vectorization extraction of remote sensing images using edge extraction algorithms. The experimental results show that the improved model has significant improvements in indicators such as Mask AP50, Box AP50, Mask mAP, and Box mAP. It can generate vectorized images of houses for the calculation of demolition costs in the pre-feasibility stage of railway construction, provide technical support for image processing work in this stage.-
Key words:
- railway engineering /
- model improvement /
- instance segmentation /
- Mask R-CNN /
- remote sensing image
-
表 1 铁路遥感影像房屋提取精度
模型 Mask AP50 Box AP50 Mask mAP Box mAP 预测用时/s 原始Mask R-CNN 0.8360 0.6703 0.8003 0.6557 11.21 改进Mask R-CNN 0.8991 0.7873 0.8539 0.7401 9.6 
下载: 导出CSV
-
[1] 国家铁路局. 铁路建设项目预可行性研究、可行性研究和设计文件编制办法:TB 10504-2018[S]. 北京:中国铁道出版社,2019. [2] 张加奇. 铁路外部环境安全隐患治理对策[J]. 中国铁路,2020(2):66-69. [3] 王 阳. 铁路建设项目征地拆迁投资控制探讨[J]. 铁路工程技术与经济,2016,31(5):22-26. [4] Hinton G E, Osindero S, Teh Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7): 1527-1554. doi: 10.1162/neco.2006.18.7.1527 [5] 韩淑梅. 基于深度学习的遥感影像铁路沿线地物检测研究[D]. 兰州:兰州交通大学,2022. [6] 高 山. 遥感技术在铁路勘察体系中的功能定位研究[J]. 铁道工程学报,2016,33(12):14-18. doi: 10.3969/j.issn.1006-2106.2016.12.004 [7] He K M, Gkioxari G, Dollár P, et al. Mask R-CNN[C]//Proceedings of 2017 IEEE International Conference on Computer Vision, 22-29 October, 2017, Venice, Italy. New York: IEEE, 2017. 2980-2988. [8] 李大军,何维龙,郭丙轩,等. 基于Mask-RCNN的建筑物目标检测算法[J]. 测绘科学,2019,44(10):172-180. [9] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 27-30 June, 2016, Las Vegas, USA. New York: IEEE, 2016. 770-778. [10] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[C]//3rd International Conference on Learning Representations, 7-9 May, 2015, San Diego, USA. ICLR, 2014.
点击查看大图
图(8) / 表(1)
