摘要
为了提高基于图像型火灾烟雾检测的准确性和实时性,提出了一种基于支持向量机的火灾烟雾检测方法。首先对图像进行增强处理,再通过累积差分法对运动区域进行检测,再基于烟雾的扩散性,对烟雾和干扰源进行初判,然后对提取的目标区域进行特征提取,最后基于支持向量机对烟雾和干扰源进行分类检测。仿真实验结果表明,该方法具有较好的鲁棒性和较高的识别率。
To improve the accuracy and real-timeof image-type smoke detection,SVM-based smoke detection method was proposed.first,the image was enhanced,moving area was detected by cumulative difference method,and the initial identification was made on smoke and interfering resource based on smoke diffusibility; then,feature extraction was made on target area; at last,smoke and interfering resource were detected separately by SVM.The simulation results showed that the method was of good robustness and high accuracy.
出处
《消防科学与技术》
CAS
北大核心
2014年第9期1052-1055,共4页
Fire Science and Technology
基金
西安科技大学博士启动基金资助项目(2014QDJ010)
陕西省教育厅自然科学基金资助项目(14JK1467)
