基于视频图像的森林火灾识别算法研究  

Study of Forest Fire Detection Method Based on Video Image

作　　者：齐怀琴[1] 徐刚 王娜[1] 

机构地区：[1]齐齐哈尔大学通信与电子工程学院,黑龙江齐齐哈尔161041

出　　处：《电视技术》2010年第9期116-118,共3页Video Engineering

基　　金：黑龙江省教育厅科研项目(10541262)

摘　　要：针对火焰的色彩特性和燃烧时的动态特性,提出一种面向视频图像序列的火灾识别算法。首先通过火焰颜色特征,在HSV模型基础上运用改进的区域增长法提取目标区域,然后根据火焰区域的重心和面积变化特性,通过设定模糊数学的隶属度函数来判定是否有火灾发生。实验结果表明,本算法实用有效,对火灾的识别率较高。In this paper, a new forest fire detection method based on video image is proposed. This method uses dynamic characteristics of flame and static characteristics. At first, through the flame color character, the HSV model and improved region-growing method are adopted to extract the target area. Then, according to the center of the flame region and the area change characteristics, whether there is fire can be judged by setting the membership function of fuzzy mathematics. The experimental results show that the method is practical and effective, and has a high recognition rate of the fire.

关 键 词：森林火灾图像 火焰识别 动态特性 区域增长法 隶属度函数 

分 类 号：TN911.74[电子电信—通信与信息系统]