摘要
针对传统火灾探测方法存在的不足,提出了一种基于支持向量机的图像型火灾探测算法,并与基于神经网络的图像型火灾探测算法做了比较。实验结果表明支持向量机克服了神经网络容易过学习、容易陷入局部极小点等不足,同时避免了人为设定特征量识别阈值时需要做大量实验和统计的复杂性。基于支持向量机的图像型火灾探测算法识别准确率高,对于小样本、高维数、非线性的分类问题效果显著。
Concerning the shortcomings of traditional fire detection,an image fire detection algorithm based on Support Vector Machine（SVM）was presented,and compared with the image fire detection based on neural network.The results show that the presented algorithm overcame the disadvantages of neural network such as over learning,being easily trapped in local minimum,etc.,and reduced the complexity of doing a lot of experiments and statistical analysis to obtain recognition threshold.The experimental results show that the image fire detection algorithm based on SVM has higher accuracy,and it is effective to solve the recognition with small samples,multi-dimension and nonlinear property.
出处
《计算机应用》
CSCD
北大核心
2010年第4期1129-1131,1140,共4页
journal of Computer Applications
基金
陕西省教育厅专项基金资助项目(08JK319)
关键词
支持向量机
图像型
火灾探测
特征提取
分类
Support Vector Machine（SVM）
image
fire detection
feature extraction
classfication
