《火灾科学（中英文）》编辑部

王莉静,郝璇,陆淞涛,丰吉科,张若轩.基于GWO-SVM多特征实时火焰识别方法[J].火灾科学,2025,34(2):126-135.

基于GWO-SVM多特征实时火焰识别方法

Multi-feature real-time flame recognition method based on GWO-SVM

DOI：10.3969/j.issn.1004-5309.2025.02.06

基金项目:天津市自然科学基金项目(20YDTPJC00840);天津城建大学研究生教改重点项目(JG-ZD-2205)

作者	单位
王莉静	1. 天津城建大学,天津,300384
郝璇	1. 天津城建大学,天津,300384
陆淞涛	1. 天津城建大学,天津,300384
丰吉科	1. 天津城建大学,天津,300384
张若轩	2. 天津农学院,天津,300380

中文关键词: 多特征提取火焰识别灰狼算法支持向量机

英文关键词:Multi-feature extraction Flame identification GWO SVM

摘要点击次数: 28

全文下载次数: 27

中文摘要:

为降低火焰干扰源引起的误报率,提出了基于GWO-SVM多特征实时火焰识别方法。首先,基于RGB颜色空间规则,结合饱和度S建立RGB-S颜色模型,对图像划分出疑似火焰区域;根据火焰独特的闪烁频率特征提取出火焰区域;基于多尺度的中等价旋转不变模式的纹理特征提取方法,获得描述图像纹理特征的特征向量。其次,采用改进的灰狼(GWO)算法对支持向量机(SVM)进行参数优化,寻找出最佳的惩罚因子和核函数参数。然后,将特征向量输入到参数优化后的SVM进行火焰分类识别。实验结果表明,本方法对火焰识别的准确率较高。

英文摘要:

To reduce the false alarm rate caused by flame interference sources, a multi-feature real-time flame identification method based on GWO-SVM is proposed. Based on RGB colour space rules and the RGB-S colour model established by saturation S, the image is divided into suspected flame areas. The flame region is extracted according to the unique flicker frequency characteristics of the flame. The feature vector describing the image texture feature is obtained based on the texture feature extraction method of multi-scale equivalent rotation-invariant patterns. The improved Grey Wolf (GWO) algorithm is used to optimize the parameters of the support vector machine (SVM) to find the best penalty factor and kernel function parameters. Then, the feature vector is input to the SVM with optimized parameters for flame classification and recognition. The experimental results demonstrate that this method achieves high accuracy in flame recognition.

关闭