科技创新与应用2025，Vol.15Issue(13)：9-13,5.DOI:10.19981/j.CN23-1581/G3.2025.13.003
基于深度学习的裂隙智能提取研究
卢宇杭 1孟治霖 1丁一 1李瀛 1翁来峰1
作者信息
- 1. 中国联合工程有限公司,杭州 310051
- 折叠
摘要
Abstract
With the rapid development of China's economic construction,engineering construction is moving in a higher,deeper and broader direction.In order to ensure the safety of project construction,it is necessary to identify the engineering geological conditions of the site before project construction.As an important part of rock mass structure,fractures have an important impact on engineering geological conditions.Therefore,extracting the structure of cracks is very important for engineering construction.Traditional crack structure extraction methods are time-consuming and labor-intensive,and have poor operability;the accuracy of crack extraction methods using traditional computer technology cannot meet engineering needs and are poor in practicality;however,there are few research on using deep learning technology to extract fractures.By studying the deep learning network of the encoder-decoder architecture,this paper determines that the image segmentation model trained by the"U-Net++"framework network and the"Resnet50"encoder-decoder network can effectively extract the fissure structure of field outcrop photos.关键词
裂隙提取/图像分割/深度学习/裂隙结构/提取方法Key words
crack extraction/image segmentation/deep learning/crack structure/extraction method分类
计算机与自动化引用本文复制引用
卢宇杭,孟治霖,丁一,李瀛,翁来峰..基于深度学习的裂隙智能提取研究[J].科技创新与应用,2025,15(13):9-13,5.
