摘要
以火灾视频和干扰视频为分析对象,利用支持向量机研究火焰及干扰物体的分类识别问题。提取火灾图像的面积变化率、圆形度以及相关系数特征,比较三种不同核函数支持向量机的训练效果,并对传统的支持向量机进行改进,提出一种自适应图像型火灾探测算法。实验结果表明:该算法能进一步提高火灾图像的分类精度和火灾识别的准确率。
Presents support vector machine to solve the classification problem based on the fire video and suspected fire video. Extracts the three features of image, which are variance ratio of flame areas, circularity and correlation coefficient. Compares with the training results of the SVMs with three types of kernels and improves the traditonal SVM algorithm. Proposes an adaptive imagetype fire detection algorithm. The experiment result shows that this algorithm can improve the classification precision of fire image and accuracy rate of fire recognization.
出处
《现代计算机》
2010年第5期68-71,共4页
Modern Computer
基金
陕西省教育厅专项基金资助项目(No.08JK319)
关键词
自适应
保角映射
数据驱动
火灾探测
支持向量机
Adaptive
Conformal Transformation
Data Driving
Fire Detection
Support Vector Machine
