摘要
针对复杂环境下火焰识别问题,提出一种基于量子引力搜索算法改进的支持向量机(SVM)的火焰识别算法。首先通过对火焰图像进行预处理并提取火焰的颜色、面积变化率和圆形度等火焰特征,并形成训练数据和验证数据;然后,采用量子引力搜索算法对SVM的核参数和惩罚因子进行最优搜索,并应用最优搜索得到的超参数建立SVM分类器模型,应用到最终的火焰图像识别中。
A flame recognition algorithm based on improved support vector machine(SVM)based on quantum gravity search algorithm is proposed aiming at the problem of flame recognition in complex environment.The algorithm firstly preprocesses the flame image and extracts the flame characteristics such as color,area change rate and circularity of the flame,and forms training data and verification data.Then,the kernel parameters and the penalty factor of the SVM which is used for optimal search are found by using the quantum gravity search algorithm,and the super-parameters obtained by the optimal search are used to establish the SVM classifier model,which is applied to the final flame image recognition.
作者
李海涛
张伟
Li Haitao;Zhang Wei(Beijing Institute of Spaceflight Test Technology,Beijing 100074,China)
出处
《电子测量技术》
2019年第18期81-84,共4页
Electronic Measurement Technology
关键词
火焰识别
量子引力搜索算法
支持向量机
flame recognition algorithm
quantum gravity search algorithm
support vector machine
