无人机森林防火系统的火灾图像识别仿真  

Fire Image Recognition Simulation of Unmanned Aerial Vehicle Forest Fire Protection System

作　　者：赵伟[1] 于芳芳 范晓婧 张南楠[1] ZHAO Wei;YU Fang-fang;FAN Xiao-jing;ZHANG Nan-nan(Northeast Forestry University,Haerbin Heilongjiang 150036,China)

机构地区：[1]东北林业大学,黑龙江哈尔滨150036

出　　处：《计算机仿真》2018年第9期459-464,共6页Computer Simulation

摘　　要：针对森林火灾识别问题,为提高图像中火灾识别效率和速率,保证图像识别的正确性,确保图像采集时间和准确性,提出一种基于无人机图像采集和改进后的BP神经网络同支持向量机(SVM)分类器结合的图像识别方法。在利用算法对火灾识别过程的基础上,以火灾特征量作为火灾识别算法的输入变量来达到减少识别训练图像和运算量,提高识别精度的目的。实验结果表明,改进后的BP神经网络同支持向量机分类器结合的图像识别方法,不仅能提高识别速率,同时也提高识别效率,减少识别误差。In order to improve the recognition rate and efficiency of fire image, to ensure the correctness of image recognition, the time and accuracy of image acquisition, an image recognition method based on Unmanned Aerial Vehicle image acquisition and an improved BP neural network combined with Support Vector Machine （SVM） classifier is proposed according to forest fire recognition problem. Based on the algorithm of fire identification process, the fire feature quantity is used as the input variable of the fire identification algorithm to reduce the recognition training image and the calculation amount, and improve the recognition accuracy. The experimental results show that the improved BP neural network combined with Support Vector Machine classifier can improve not only the recognition rate, but also the recognition efficiency, and reduce the recognition error.

关 键 词：无人机 森林防火 神经网络 支持向量机分类器 火灾识别 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]