基于红外成像多特征联合的乙烯气体泄漏检测  

Ethylene gas leaking detection based on combination of multi-features of infrared imaging

作　　者：王建平 李俊山[1] 杨亚威 胡双演[1] 孙胜永 

机构地区：[1]第二炮兵工程大学信息工程系,陕西西安710025

出　　处：《传感器与微系统》2015年第6期119-122,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(61175120)

摘　　要：针对目前乙烯气体泄漏检测存在效率低、精度差以及无法准确标定泄漏位置等问题,提出了一种基于红外成像多特征联合的乙烯气体泄漏检测方法。该方法首先利用红外相机获取生产车间的红外视频图像,通过联合帧间差分和背景差分的方法获取疑似区域;其次对疑似区域进行泄漏气体成像的统计特征和小波能量变化特征分析,找出特征阈值;最后将多个特征阈值条件联合在一起,作出是否存在泄漏的判定。实验结果表明:该方法准确、高效且具备实时性,其泄漏检测的准确率达到95.6%。Aiming at the problems of low efficiency,precision,and that the leak location cannot be accurately calibrated in ethylene gas leakage detection,present a method for detecting ethylene gas leakage based on multiple features combination of infrared imaging. Firstly,Infrared image of the production workshop is obtained by the infrared camera,and the suspected area is obtained by the method of the combination of frame difference and background difference;secondly,the statistics features and the wavelet energy variation characteristics of gas leakage imaging in suspected area is analyzed to find out the feature threshold;finally,the multiple feature threshold conditions is combined together,to determine whether there is leakage. The experimental results show that this method is accurate,efficient and real-time,and the leak detection accuracy rate reaches 95. 6%.

关 键 词：红外成像 气体检测 统计特征 能量变化特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]