基于改进双流卷积网络的火灾图像特征提取方法  

Fire Images Features Extraction Based on Improved Two-stream Convolution Network

作　　者：徐登 黄晓东[3] XU Deng;HUANG Xiao-dong(Jiangsu Internet of Things and Manufacturing Information Engineering Research Center,Changzhou,Jiangsu 213164,China;School of Information Engineering,Changzhou Vocational Institute of Mechatronic Technology,Changzhou,Jiangsu 213164,China;Key Laboratory of MEMS of Ministry of Education,Southeast University,Nanjing 210096,China)

机构地区：[1]江苏省物联网与制造业信息化工程技术研究开发中心,江苏常州213164 [2]常州机电职业技术学院信息工程学院,江苏常州213164 [3]东南大学MEMS教育部重点实验室,南京210096

出　　处：《计算机科学》2019年第11期291-296,共6页Computer Science

基　　金：国家自然科学基金项目(61604040)资助

摘　　要：基于图像处理技术的火灾监测,是近年来火灾监控领域的重要分支。对于开阔场景的火灾监测,利用火灾发生时产生的烟雾和火焰的动、静特性,以双流(Two-Stream)卷积神经网络作为理论基础对火灾进行检测识别。双流卷积神经网络采用空间流与时序流分别提取视频中的空间信息与时序信息,然而火灾初期的信息较为微弱,特征不够明显。为进一步提高初期的识别率,提出一种空间增强网络作为双流卷积神经网络的空间流来提取并增强视频的空间信息。空间增强网络同时对当前帧图片V t和上一帧图片V t-1做卷积,用V t的卷积特征与V t-1的卷积特征做减法,保留卷积特征差异性,再将卷积特征差与当前帧V t的卷积特征相加,从而增强对V t的空间特征卷积;双流卷积网络的时间卷积流对当前帧的光流图片V t′进行时序特征卷积;最后将增强后的空间特征与时序特征融合进行分类。实验结果表明,改进后的双流卷积网络的识别率比原始的双流卷积网络提高了6.2%,且在公开数据集上的测试准确率达到了92.15%,从而证明了该方法的有效性和优越性。此外,与其他方法相比,该网络具有低深度、高识别率的特征,不仅能提高火灾和烟雾的识别率,而且实现了火灾的早期发现,缩短了检测时间。Fire detection based on image processing technology is an important branch in the field of fire monitoring in recent years.Aiming at the fire detection of open environment,using the dynamic and static characteristics of smoke and flame generated during the fire,the two-stream convolutional neural network is used as the theoretical basis to detect the fire.The two-stream convolutional neural network uses spatial and temporal streams to extract spatial information and temporal information in the video respectively.However,the information in the early stage of the fire is weak and the features are not obvious enough.In order to improve the initial recognition rate,a spatial enhancement network was proposed as the spatial stream of the two-stream convolutional neural network to extract and enhance the spatial information of the video.The spatial enhancement network simultaneously convolves the current frame V t and the previous frame V t-1,subtracting the convolution features of the V t image with the convolution features of the V t-1 image,to preserve the difference of the convolution features,and adding the convolution features difference to the convolution features of the current frame V t,thereby enhance the spatial features convolution of the current frame V t-1.Temporal stream of two-stream convolutional network convolves the optical flow image V t′of the current frame to get the temporal features.Finally,the enhanced spatial and temporal features are fused to classify.The experimental results show that the improved two-stream convolutional network has a 6.2%higher recognition rate than the original two-stream convolutional network,and achieved 92.15%recognition rate on the public dataset,indicating the effectiveness and superiority of the proposed method.Comparing with other methods,the network structure is designed lower but achieves good results,improves the identification accuracy of fire and smoke as well as realizes the early warning of fire,shorten detection time.

关 键 词：开阔空间火灾监测 空间特征增强网络 双流卷积神经网络 时空特征融合 光流法 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]