摘要
烟雾检测技术在火灾早期预警阶段发挥着重要的作用,准确快速地烟雾检测算法具有非常重要的实际应用价值。针对现存的烟雾检测算法抗干扰能力差、实时性不强、复杂场景下适应性差的问题,本文提出基于多特征融合的自适应烟雾检测算法。首先,算法通过改进的三帧差分法提取基于图像块的运动区域;其次,提取烟雾图像块的HSV颜色特征、纹理特征、能量特征、面积变化率和LBP特征,多特征融合之后通过支持向量机(SVM)算法训练烟雾检测模型,进行烟雾检测。为了有效评估基于多特征自适应的烟雾检测模型的有效性,在复杂的烟雾场景中进行试验,试验结果证明该烟雾检测算法具有良好的鲁棒性。
Smoke detection technology plays an important role in the early warning stage of fire.Accurate and rapid smoke detection algorithm is of great practical application value.The existing smoke detection algorithm has poor anti-interference ability,low real-time performance and poor adaptability in complex scenes.In order to solve the above problems,an adaptive smoke detection algorithm based on multi-feature fusion is proposed.Firstly,the algorithm detects the motion region based on image block by the improved three-frame difference method.Secondly,HSV color feature,texture feature,energy feature,area change rate and LBP feature of the smoke image block are extracted.After multi-feature fusion,the support vector machine(SVM)is adopted to train the smoke detection model.To effectively evaluate the smoke detection model based on multi-feature adaptive,the experiments are carried out in complex smoke scenarios.The experimental results show that the proposed smoke detection algorithm has great robustness.
作者
殷梦霞
王理
孙连营
Yin Mengxia;Wang Li;Sun Lianying(China Academy of Building Research, Beijing 100013, china)
出处
《建筑科学》
CSCD
北大核心
2019年第9期26-31,共6页
Building Science
关键词
烟雾检测
三帧差分法
支持向量机
多特征自适应
smoke detection
three-frame difference method
SVM
adaptive multi-feature fusion
