摘要
为了确保火灾探测结果的可靠性和准确性,从林火燃烧时火焰和烟雾特征出发,对现有的林火探测技术进行了分析,提出了一种基于K-means和颜色模型的林火辨识方法。首先使用Kmeans算法对采集到的彩色图像进行分割,根据火灾发生时火焰和烟雾的颜色特征,采取一种改进的颜色模型对分割出来的子图像进行辨识,对疑似火焰子图像和疑似烟雾子图像进行初步确认,然后从疑似子图像中提取出火焰和烟雾的特征输入到RBF神经网络,判断是否确实发生火灾。
In order to ensure the reliability and accuracy of detection resuhs, from flame and smoke characteristics of the for- est fires, the characteristics of the existing forest fire detection technology are analyzed. In order to solve these defects, a kind of forest fire recognition method based on clustering algorithm and color model is proposed. First, K-means algorithm is used to seg- ment color images which were collected, according to the color characteristics of flame and smoke, the sub images of segmentation are preliminarily identified with an improved color model,then the characteristics of flame and smoke from the suspected sub images are inputted into RBF neural network to determine whether there is a fire. The simulation experiments show that the method can ef- fectively identify on forest fire, providing powerful basis for subsequent fire fighting work.
出处
《电子技术应用》
北大核心
2015年第2期163-166,共4页
Application of Electronic Technique
基金
陕西省西安市未央科技区项目(2012-03)
关键词
森林火灾
火焰
烟雾
辨识方法
forest fire
flame
smoke
identification method
