GAO Feng-wei,WEI Wei,CHENG Yang.Resarch for Early Wildfire Smoke Video Detection[J].Journal of Chengdu University of Information Technology,2018,(05):509-516.[doi:10.16836/j.cnki.jcuit.2018.05.005 ]
野外早期火灾烟雾视频检测技术研究
- Title:
- Resarch for Early Wildfire Smoke Video Detection
- 文章编号:
- 2096-1618(2018)05-0509-08
- 关键词:
- 视频烟雾检测; 卷积神经网络(CNN); 线性动态系统(LDS); 动态纹理
- Keywords:
- video smoke detection; Convolutional Neural Network(CNN); Linear Dynamic Systems(LDS); dynamic textures
- 分类号:
- TP391.41
- 文献标志码:
- A
- 摘要:
- 针对视频中野外早期火灾烟雾形状模糊,颜色相对背景对比不明显,外观特性多样等的特点,提出一种并行的基于深度学习和动态纹理特征的烟雾识别方法。实验中,针对视频烟雾检测中视频帧的噪声,采用均值滤波的方法对视频帧进行过滤。并给出一种基于野外早期火灾烟雾视频的深度卷积神经网络结构,并对该结构进行训练和测试。针对烟雾的动态纹理特征,提出建立线性动态系统模型,结合支撑向量机对烟雾视频帧进行分类识别。最后结合卷积神经网络的识别结果和动态纹理特征识别结果提出一种混合矩阵的判定方法。实验结果表明,方法相对于传统烟雾识别方法在早期火灾烟雾的检测上有更高的准确率。
- Abstract:
- A smoke recognition method of deep learning and dynamic texture features based on parallel for the early fire smoke video's blurry shape and the color contrast is not obvious compared with the background, and the appearance characteristics are varied.In the experiment,using the method of mean filter for filtering the video frame.And gives a convolutional neural network structure for early fire smoke based on video,then training and testing of the structure.For the dynamic texture features of smoke,making a linear dynamic system model and detecting the video frames based on support vector machine.The smoke is proposed based on a judgment method of the mixing matrix convolution neural network recognition results and dynamic texture recognition results.The experimental results show that this method compared with the traditional smoke recognition method has higher accuracy in the detection of early fire smoke.
参考文献/References:
[1] 刘嘉,汪则灵.浅谈火灾危害与环境保护[J].经营管理者,2010,17:180-181.
[2] etin A E,Dimitropoulos K,Gouverneur B,et al.Video fire detection-Review[J].Digital Signal Processing,2013,23(6):1827-1843.
[3] Ye Wei,Zhao J,Wang S,et al.Dynamic texture based smoke detection using Surfacelet transf-orm and HMT model[J].Fire Safety Journal.2015,73:91-101.
[4] Chunyu Yu,Zhibin Mei,Xi Zhang.A real-time video fire flame and smoke detection algorithm[J].Procedia Engineering.2013,62:891-898.
[5] 胡燕,王慧琴,姚太伟,等.基于Harris特征点检测与跟踪的火灾烟雾识别[J].计算机工程与应用,2014,50(21):180-183.
[6] ByoungChul Ko,JunOh Park,Jae-Yeal Nam.Spati-otemporal bag of features for early wild fire smoke detection[J].Image and Vision Computing,2013,31:786-795.
[7] Doretto G,Chiuso A,Wu Y N,et al.Dynamic Textures.[J].International Journal of ComputerVision,2003,51(2):91-109.
[8] Coviello E,Mumtaz A,Chan A B,et al.Growinga bag of systems tree for fast and accurate classification[C].Computer Vision and Pattern Recognition.IEEE,2012:1979-1986.
[9] Ravichandran A,Chaudhry R,Vidal R.Categorizing Dynamic Textures Using a Bag of DynamicalSystems[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2013,35(2):342-353.
[10] Mumtaz A,Coviello E,Lanckriet G R G,et al.A Scalable and Accurate Descriptor for DynamicTextures Using Bag of System Trees[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2015,37(4):697-712.
[11] Barmpoutis P,Dimitropoulos K,Grammalidis N.Smoke detection using spatiotemporal analysis,motion modeling and dynamic texture recogniti-on[C].Signal Processing Conference.IEEE,2014:1078-1082.
[12] Dimitropoulos,Kosmas,Barmpoutis,Panagiotis,et al.Nikos.Higer order linear dynamical systems for smoke detection in video surveillanceapplications[J].IEEE Transactions on Circuitsand Systems for Video Technology,2016,8215(c):1-13.
[13] Chongyuan Tao,Jian Zhang,Pan Wang,Smoke d-etection based on deep convolutional neural networks[C].IECON 2016-42nd Annual Conference of the IEEE Industrial Electronics Society,2016:150-153.
[14] 陈俊周,汪子杰,陈洪瀚,等.基于级联卷积神经网络的视频动态烟雾检测[J].电子科技大学学报,2016(6):992-996.
[15] Hinton G E,Srivastava N,Krizhevsky A,et al.Improving neural networks by preventing coadaptation of feature detectors[J].Computer Scie-nce,2012,3(4):212-223.
[16] Kingma D P,Ba J.Adam:A Method for Stochas-tic Optimization[J].Computer Science,2014.
[17] 朱旭阳,李思昆.动态纹理合成技术研究综述[J].系统仿真学报,2001(S2):(23-25)+38.
[18] P V Overschee,B D Moor.N4SID:Subspace Algorithms for the Identification of Combi-ned Deterministic-Stochastic Systems[J].Automatica,1994,30:75-93.
[19] R.Shumway and D.Stoffer.An Approach to Time Series Smoothing and Forecasting Using the EM Algorithm[J].Time Series Analysis.1982,3(4):253-264.
[20] 唐杰,周洋,杨萌,等.采用颜色混合模型和特征组合的视频烟雾检测[J].光电子·激光,2017,28(7):751-758.
备注/Memo
收稿日期:2017-10-14 基金项目:四川省教育厅重点科研资助项目(2017Z26) 基金项目:四川省教育厅重点科研资助项目(2017Z26) 基金项目:国家自然科学基金资助项目( 41375043、41405030、41505031)