【摘要】 针对现有条件下的岩屑录井中岩屑识别率低、识别速度慢等问题,从特征提取和分类器方面对岩屑岩性识别进行了分析研究.采用二级分类器的思想,首先通过颜色特征和和差直方图特征采用朴素贝叶斯分类器将岩屑粗分为泥岩和砂岩,然后进一步采用贝叶斯分类器,通过颜色特征和和差直方图特征分别将泥岩和砂岩进行进一步的细分.实验结果表明,粗分的识别率、泥岩细分的识别率和砂岩细分的识别率分别能达到94.79%、97.59%和90.28%.这种识别方法更加符合现实的应用需求,有着更高的识别率,为岩屑岩性分析工作提供了可靠的依据.更多还原

【Abstract】 In current cutting logging,low recognition rate and identifying slow are the main existing problems.In order to solve these problems,some studies about cuttings identification has been made from the feature extraction and classification.Within the idea of the two categories,first via color and the Sum and Difference Histograms feature,the cuttings will be roughly divided into mudstone and sandstone by Native Bayes classifier.Then the mudstone and sandstone independent via color and the Sum and Difference Histogram feature subdivided with classifier.The experimental show that the accuracy about 94.79%for roughly division and 97.59%for mudstone subdivision,90.28%for sandstone subdivision.This identification method is suitable for the application requirements and has a higher recognition rate,and it will provide a reliable basis for the further cuttings identification.更多还原