摘要
在所有测井资料中,成像测井资料是最直观有效的裂缝识别资料。裂缝拾取的自动化可有效减少人力消耗。采用计算机视觉领域的图像语义分割模型DeepLabv3+,对裂缝区域进行像素分割,在分割结果的基础上利用霍夫变换提取裂缝形态参数。与传统方法相比,该方法能够更为准确,快速地提取裂缝区域,求取裂缝产状。
Imaging logging data is the most intuitive and effective for fracture identification.Automatic fracture identification effectively reduces human cost.In this paper,DeepLabv3+,the most advanced image semantic segmentation model in the field of computer vision,is used to segment the fracture zone by pixels.Based on segmented results,Hough transform is used to extract fracture shape parameters.Compared with a traditional method,this method can more accurately and quickly extract fracture zones and describe fracture shapes.
作者
李冰涛
王志章
孔垂显
蒋庆平
王伟方
雷祥辉
LI Bingtao;WANG Zhizhang;KONG Chuixian;JIANG Qingping;WANG Weifang;LEI Xianghui(China University of Petroleum,Beijing 100000,China;Research Institute of Exploration& Development,PetroChina Xinjiang Oilfield Company,Karamay,Xinjiang 834000,China)
出处
《测井技术》
CAS
2019年第3期257-262,共6页
Well Logging Technology
关键词
成像测井
裂缝参数
语义分割
霍夫变换
卷积神经网络
imaging logging
fracture parameters
semantic segmentation
Hough transform
convolution neural network
