摘要
提出了一种基于火焰图像动态特征的火灾识别算法。首先在HSI空间对捕捉的图像进行彩色分割,通过相邻两帧图像判断视频中是否存在可疑火焰。当有可疑火焰存在时,再连续捕捉多帧图像,进行火焰中心定位并提取火焰动态特征,最后判断是否发生火灾。实验结果表明,该算法具有比较好的鲁棒性,能够有效地提高火灾识别的准确率、降低误报漏报率。
A fire detection method based on flame dynamic features is proposed. At first, from two adjacent frames, color segmentation and frame difference are applied to judge whether there is suspicious fire in the video or not. When there is, flame dynamic features are extracted from the following frames to determine the presence of fire or non-fire patterns. Experimental results show that the method can detect fire successfully in video.
出处
《测控技术》
CSCD
2007年第5期7-9,共3页
Measurement & Control Technology
关键词
火灾探测
火焰动态特征
彩色分割
fire detection
flame dynamic features
color segmentation
