摘要
本文采取了红外图像特征提取的方法检测铜电解生产系统中阴阳极板间的短路.首先采集烧板和非烧板的样本;然后在局部二值模式LBP的基础上提出差分LBP特征检测算子;最后选用支持向量机SVM分类器进行分类来提高识别率.实验表明,通过红外图像检测短路的方法可行,且改进的D-LBP特征比LBP特征及其他传统特征提取算法更适合烧板和非烧板的检测,在所选取的5000张样本中其检测正确率达到了98.5%.
In this paper,the method of infrared image feature extraction is adopted to detect the short circuit of anode and cathode in copper electrolysis production system.First of all,samples of the burning plate and the non-burning plate are collected,then differential LBP feature detection operator based on the LBP is put forward.Finally,the SVM classifier is used to improve the recognition rate.The experiment result shows that the method of detecting burning plate by infrared image is feasible,and the improved differential LBP features are more suitable for the classification of burning plate and non-burning plate than the LBP and other traditional feature extraction algorithms.Of the 5000 selected samples,the correct rate is 98.5%in the tests.
出处
《北方工业大学学报》
2016年第3期1-7,共7页
Journal of North China University of Technology
关键词
铜电解
红外图像
LBP
烧板检测
SVM
copper electrolysis
infrared image
LBP
burning plate detection
SVM
