摘要
为了提高初至波自动拾取的稳定性和高效性,减少初至波拾取耗费的人力,将当前迅猛发展的深度学习技术应用到地震数据处理领域。把初至拾取看作是二分类问题,初至波为一类,背景为另一类,参考全卷积神经网络在图像语义分割和边缘检测的成功应用,提出了利用全卷积神经网络来拾取初至波的新算法。通过搭建三个不同深度的全卷积神经网络结构并对其进行性能测试,选取性能最优的网络,并与商业地震数据处理软件Tomo Plus自动拾取的结果进行比较,在高噪声情况下取得了更好的结果,验证了利用全卷积神经网络拾取初至波的可行性。
To improve the stability and efficiency of the first arrival pickup and reduce the human resources,this paper applies the current rapidly developing deep learning technology to the field of seismic data processing. The first arrival pickup is regarded as the two-classification problem,the first arrival is one kind,and the background is another class.With the successful application of fully convolutional networks （FCN） in image semantic segmentation and edge detection,this paper proposes a new algorithm which using FCN to pick up the first arrival.Through testing three networks of different depths, the network structure with optimal performance is selected and compared with the result obtained by the commercial software of seismic data processing TomoPlus.It gets better result under high noise which verifies that it is feasible to pick up the seismic first arrival by using FCN.
作者
刘佳楠
武杰
Liu Jianan;Wu Jie(Department of Modern Physics,University of Science and Technology of China,Hefei 230026,China;State Key Laboratory of Particle Detection ＆ Electronics,University of Science and Technology of China,Hefei 230026,China)
出处
《信息技术与网络安全》
2018年第11期58-63,共6页
Information Technology and Network Security
基金
国家自然科学基金项目(41574106)
国家科技重大专项(2017ZX05008-008-041)
国家重大科研装备研制项目(ZDYZ2012-1-05-03)
关键词
海量数据
初至波拾取
深度学习
全卷积神经网络
massive data
first arrival picking
deep learning
fully convolutional networks
