基于混合高斯背景模型的SF6泄漏自动检测  

Automatic SF_6 leakage detection based on Gaussian mixture background model

作　　者：蔺丽华[1] 吴冬梅[1] 李杰[1] 刘健[1,2] 

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054 [2]陕西电力科学研究院,陕西西安710054

出　　处：《西北大学学报（自然科学版）》2014年第3期379-382,共4页Journal of Northwest University（Natural Science Edition）

基　　金：陕西省工业攻关计划基金资助项目(2011K09-36);陕西省教育厅科研计划基金资助项目(12JK0528);陕西省科技攻关计划基金资助项目(2012K06-16);陕西省教育厅专项科研计划基金资助项目(08JK356)

摘　　要：为了能够针对可能发生的SF6泄漏情况,进行连续及时的无人监测,基于红外辐射成像技术提出一种自动检漏方法。利用SF6泄漏在红外视频中类似烟雾的动态特性,采用混合高斯背景建模的方法进行SF6气体的泄漏检测,并基于数学形态学方法去除斑点等噪声,完成对泄漏区域的自动标记定位。通过对多个红外气体成像检漏仪采集到的SF6视频进行泄漏检测,实验结果表明,该方法能克服光线变化等背景扰动影响,实现泄漏点的检测和定位。A method, based on the infrared gas imaging, was proposed for automatic continuous monitoring and detecting the leak of SF6 gas. This new method was developed by integrating three technologies, including visualizing the invisible SF6 gas cloud as a standard video image on TV monitor through the backscatter absorp- tion gas imaging; subtracting background noise signals using Gaussian mixture model, and removing small spots in the image by an algorithms of mathematical morphology. A SF6 gas detecting system was designed based on this method, and analysis of the images collected by this system in a field test revealed that this new method enable to detect leaked SF6 gas and locate the leaking spots effectively with no interference from light and other environment factors.

关 键 词：高压电气设备 SF6气体 红外视频图像 泄漏检测 混合高斯背景模型 

分 类 号：TM63[电气工程—电力系统及自动化]