基于图像处理与神经网络的岩石组构识别  

Research of rock texture identification based on image processing and neural network

作　　者：程国建[1] 马微[1] 魏新善[2] 荣春龙[2] 南珺祥[2] 

机构地区：[1]西安石油大学计算机学院,陕西西安710065 [2]中国石油长庆油田分公司勘探开发研究院,陕西西安710021

出　　处：《西安石油大学学报（自然科学版）》2013年第5期105-110,1,共6页Journal of Xi’an Shiyou University（Natural Science Edition）

基　　金：国家自然科学基金项目(编号:40872087)

摘　　要：针对岩石薄片组构鉴定的复杂性问题,基于岩石薄片数字图像的数值分析,提出一个自动进行岩石组构识别分类的系统方法:使用图像处理技术对岩石薄片的灰度数字图像进行特征提取,将其作为神经网络的输入,网络输出是图像对应的岩石组构类别;使用来自鄂尔多斯盆地苏里格地区的100幅岩石薄片图像进行实验,神经网络的训练集、验证集和测试集随机分为70、15和15幅图像,测试结果表明自动识别岩石组构的正确率达93.3%.该方法对于地学问题具有较好的综合判别能力,可以快速和准确地进行岩石组构识别.For the complexity of rock microstructure identification,an automatic classification and identification method of rock texture is proposed based on analysis of the digital image of rock thin section. The features of the digital grey images of the rock thin sections are extracted using image processing technique in a neural network toolbox,and then the features are as input of a neural network model. The output of the neural network model is the classification of rock microstructure. The experiment is carried out using 100images of rock thin-sections from Sulige region in Ordos Basin,and the whole dataset of 100 images is randomly divided into a training dataset of 70 images,a validation dataset of 15 images and a testing dataset of 15 images. The experimental results show that the correct rate of the rock microstructure classification and identification method can reach to 93. 3%. Therefore,the proposed method i can quickly and accurately carry out the classification and identification of rock microstructure.

关 键 词：图像处理 神经网络 岩石组构识别与分类 

分 类 号：P618.13[天文地球—矿床学] TP183[天文地球—地质学]