基于ViBe模型的VOCs泄漏自动识别算法

VOCs Leakage Automatic Identification Algorithm Based on ViBe Model

挥发性有机物(Volatile Organic Compounds,VOCs)是大气污染的重要来源,也是臭氧和有机颗粒物的重要前体物质,对人体和环境都有很大的伤害。LDAR(Leak Detection and Repair)技术通过定期检修设备与管阀件等VOCs主要泄漏源,对石化行业降本减排具有重要意义。针对当下红外摄像仪检测VOCs时,完全依赖人眼辨识泄漏发生与否的缺陷,提出基于简单且运算快速的Visual Background extractor(ViBe)模型的VOCs泄漏自主辨识算法。为提高模型环境适应性,模型结合二值化和图像形态学开运算处理,以及K-means++聚类算法,实现在原视频帧中框选出泄漏位置,为人工辨识泄漏提供帮助和参考。

Volatile Organic Compounds(VOCs)are important sources of air pollution,precursor substances of ozone and organic particulates,which have great harm to human body and the environment.Leak Detection and Repair(LDAR)technology is of great significance for reducing costs and emissions in the petrochemical industry through regular overhauling main VOCs leakage sources such as equipment and pipe valves.In view of the fact that VOCs leakage,detected by current infrared camera,completely depends on human eyes to identify whether the leakage occurs or not,an autonomous identification algorithm is proposed for VOCs leakage based on the simple and fast model called Vi Be(Visual Background Extractor).The method is combined with binarization,image morphological processing,and k-means++clustering algorithm to improve environment adaptability.It achieves the goal of box selection of the leakage location in the original video,which provides some help and references for manual identification of the leakage.