基于CLBP和支持向量诱导字典学习的煤岩识别方法

发布时间：2018-06-07 21:01

本文选题：煤岩识别 + CLBP　；参考：《煤炭学报》2017年12期

【摘要】：针对现有煤岩识别方法在训练样本不充足情况下的识别效果普遍不太理想这一情况,提出了一种基于完备局部二值模式(CLBP)和支持向量诱导字典学习的煤岩识别方法。该方法分4大步完成:(1)提取煤岩图像的多尺度CLBP特征向量;(2)对训练样本的CLBP特征向量进行支持向量诱导字典学习,得到一组煤岩表征字典、煤岩类别权向量和偏移量;(3)计算测试样本在煤岩表征字典上的表示即编码向量;(4)采用判别函数完成测试样本编码向量的类别判定。结果表明:与现有其他常用方法相比,所提出方法有着更高的正确识别率,特别是在训练样本不充分的随机抽样实验条件下,其正确识别率仍然很高;耗时的字典学习并没有影响到所提出方法的实时性;所提出方法占用的存储量不受训练样本数量的制约,这在一定程度上为将来硬件实现带来了便利。
[Abstract]:A coal rock recognition method based on complete local two value model (CLBP) and support vector induction dictionary learning is proposed to identify coal rock recognition methods under insufficient training samples. This method is completed in 4 major steps: (1) extracting multi-scale CLBP eigenvectors of coal and rock images; (2) The CLBP eigenvector of the training sample is carried out by support vector induction dictionary learning, and a group of coal rock characterization dictionaries, coal rock category weight vectors and offset are obtained. (3) the representation of the test samples in the coal rock representation dictionary is the encoding vector; (4) the discrimination function is used to complete the classification of the test sample coding vector. Compared with the common methods, the proposed method has a higher correct recognition rate, especially in the condition that the training sample is not sufficient, the correct recognition rate is still high; the time-consuming dictionary learning does not affect the real-time performance of the proposed method; the proposed method is not restricted by the number of training samples. To a certain extent, it will bring convenience to the realization of hardware in the future.
【作者单位】：中国矿业大学(北京)机电与信息工程学院;
【基金】：国家重点研发计划资助项目(2016YFC0801800) 中央高校基本科研业务费专项资金资助项目(2010YJ06)
【分类号】：TD67

【相似文献】