装甲车辆动力舱火焰视频图像多特征探测与识别方法

Video Image Detection and Recognition Method Based on Multi-Feature Flame in Engine Cabin of Armored Vehicle

针对当前装甲车辆动力舱线式火焰传感器探测火焰响应时效性差问题,提出了基于多特征的装甲车辆动力舱火焰视频图像探测和识别方法;通过在动力舱安装视频采集火焰视频,对火焰固有多特征(静态特征和动态特征)进行分析,首先对火焰的RGB和HSI模型颜色分量特征、形状特征等静态特征进行分析确定疑似火焰区域,其次通过提取疑似火焰区域闪烁特征、面积变化特征等动态特征以及连通区域像素点总数设置经验阈值进行算法分析;通过建立动力舱模拟环境实验室并验证,多特征火焰分析和识别算法能快速确定动力舱的火焰发生,满足火焰探测实时性要求。

Aiming at the problem of the poor response time of flame detection by the wire flame sensor in engine cabin of armored vehicle,a video image detection and recognition method based on multi-feature flame in engine cabin of armored vehicle was proposed.By installing video in engine cabin to collect video of flame,the inherent multiple characteristics of flame(static characteristics and dynamic characteristics)were analyzed.The color component characteristics,shape characteristics static characteristics of the RGB and HSI models of the flame were analyzed to extracted the suspected flame area.The algorithm was analyzed by extracting dynamic features such as flicker feature and area change feature of suspected flame region and setting empirical threshold of total number of pixels in connected region.The multi-feature flame analysis and recognition algorithm can quickly determine the flame occurrence of the power cabin and meet the real-time requirement of flame detection by establishing the simulation environment laboratory of the power cabin and testing.