一种基于支持向量机与纹理的岩屑识别算法  

A Cutting Recognition Algorithm Based on Support Vector Machine and Texture

作　　者：杨晓敏[1] 冉飞[2] 吴炜[1] 陈默[1] 何小海[1] 

机构地区：[1]四川大学电子信息学院图像信息研究所,四川成都610064 [2]中石化石油工程西南公司地质录井分公司研究院,四川绵阳621000

出　　处：《石油化工高等学校学报》2009年第2期88-92,共5页Journal of Petrochemical Universities

基　　金：四川省科技攻关项目资助(No.05GG021-026-03)

摘　　要：针对在随钻过程中如何自动、快速、准确地识别岩屑的问题,对岩屑的纹理进行了深入的分析研究,采用Gabor滤波器对不同岩性的岩屑图像进行纹理提取和分析,提取了反映不同岩屑图像纹理结构的特征参数,最后将支持向量机应用到岩屑的识别过程中,取得了较好的识别效果。通过对四川洛带气田4口井的现场随钻,结果证明算法识别正确率达到90%以上,为快速自动录井提供了一条有效的途径。In the cutting logging process, how to recognize the cuttings automatically,quickly and correctly is very important. A cutting recognition algorithm based on Gabor filter and SVM was proposed to resolve this problem. The algorithm used Gabor filter to get the cuttings＇ features. Then the features were used to train the support vector machine classifier and the cuttings were classified by using the trained support vector machine. Using the algorithm, after logged 4 wells of Luodai gas reservoir, a high recognition rate can be reached at 90%. The experimental results demonstrate the algorithm is feasible, robust and applicable.

关 键 词：岩屑识别 支持向量机 纹理 GABOR滤波器 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]