摘要
针对在PDC钻头条件下石油录井过程中的岩屑岩性难以用肉眼识别,以及人工分类速度较慢的问题,该文利用傅立叶周向谱能量法原理,将岩屑图像在频域中按照频率由低到高对周向谱能量进行分级,将各级能量占总能量的比率作为该图像的特征量,然后应用Bayes分类器对其进行分类处理,将所得结果与人工分类结果做比较,并验证了方法的可靠性。实验证明,这种方法能更好的分析岩屑的岩性,对泥岩的分类准确率高达99.44%。
In the process of logging, it＇s quite slow for human eyes to identify the cuttings made by the PDC bit. In order to extracting characteristics of cuttings＂ image texture effectively, using the algorithm of circular Fourier spectral energy, the image＇s spectral energy in different frequency ranges is classified. Hence the distribution ratio of the spectral energy is calculated and taken as the characteristic. With the basis of the results of artificial identifying, the characteristics of the standard samples are taken as the training characteristics to test the ones of the random samples using Bayes classifier in the identifying process. The reliability of the method is also tested. It proved that the method was relatively good to analyze the character of cuttings with an identifying accuracy 99.44% for mudstone.
出处
《石油实验地质》
CAS
CSCD
北大核心
2008年第4期420-423,共4页
Petroleum Geology & Experiment
基金
国家高技术研究发展计划(863计划)项目(2002AA615170)
