基于全卷积神经网络的遥感图像线性构造解译方法——以云县官房铜矿区为例

Linear structure interpretation method of remote sensing image based on full convolution neural network:An example of Guanfang copper mining area in Yunxian County

文章研究了深度学习方法在地质构造解译中的应用,探究了相比传统的线性构造方法更为高效且无需先验知识的方法。以基于全卷积神经网络(FCN)的图像像素注释方法实现了遥感数据对于线性构造解译半自动解译。选择云南省云县官房铜矿矿区作为实验区域,绘制的图件表明该解译方法能够满足普通地质研究的基本需求,同时也能作为人工线性构造解译工作的初步参考,具有一定的研究意义。而与其他传统自动解译方法对比,可以发现在解译精度、效率和可重复使用性上都存在一定的优势。这些研究成果对于地质构造解译的自动化发展具有重要的参考价值,也为遥感解译智能化的发展提供了新思路。

This paper investigates the application of deep learning method in geological structure interpretation and explores for a more efficient and prior-knowledge-free approach compared to traditional linear construction method.An image pixel annotation method based on fully convolution neural network(FCN)is used to achieve semi-automated interpretation of linear structure with remote sensing data.Guanfang copper mining area in Yun County,Yunnan Province is selected as the experimental area,and the study result indicates that this interpretation method can meet the basic need of general geological research and serves as a preliminary reference for manual linear interpretation work with a certain research significance.Compared with other traditional automatic interpretation methods,it has the advantage in interpretation accuracy,efficiency and reusability.These research results have an important reference value for the automation development of geological structure interpretation,and it also provides a new idea for the intelligent development of remote sensing interpretation.