摘要
介绍了基于红外视觉的船舶机舱火灾探测方法.首先将滑动时间窗内的若干帧红外火焰图像叠加到一个投影平面上,投影图像像素的灰度值反映了火焰的空间分布,采用投影矩来描述火焰的动态特征.然后,基于目标的投影矩特征,采用支持向量机来区分火焰和非火焰目标.实验中采用了两种火焰和两种疑似目标视频,实验结果表明,该方法具有较高的实时性和良好火焰探测效果.
A fire detection method of marine engine room based on infrared vision is introduced First, we project the infrared flame images in the sliding windows onto a projection plane. The grey values of pixels in the projection image signify the spatial distribution of the flame. We use the invariant moments of the projection image describe the dynamic distribution of the flame. Second, we use support vector machines to distinguish fame object from fake object based on the projection moments. In experiments, we adopted two kinds of flames and two kinds of fake targets. Experimental results show that the algorithm can be used for kinds of flame objects detection effectively, with high real-time and robustness performance.
出处
《船舶工程》
CSCD
北大核心
2009年第3期75-78,共4页
Ship Engineering
基金
国家863课题(2007AA1Z158)
关键词
船舶机舱
火焰检测
红外视觉
投影矩
marine engine room
flame monitoring
infrared vision
projection moment
