摘要
提出了一种基于堆叠深度卷积沙漏网络的步态识别方法。为了解决人体建模中关节点准确定位的问题,采用基于深度卷积的沙漏网络来提取步态图上的关节点坐标,并计算肘关节与膝关节的角度作为运动特征。为了解决行走速度变化带来的影响,采用动态时间规整(Dynamic Time Warping)对特征序列进行距离计算。通过最近邻分类器对结果进行准确分类。该方法在公共CASIA-B数据集与TUM-GAID数据集上进行了验证并与其他方法进行比较,结果表明该方法有较高的识别率。
A new approach for gait recognition is proposed based on stacked depth convolution hourglass network. In order to locate the joints in human modeling accurately, the convolution hourglass network based on deep convolution network is used to extract the joint coordinates of gait image, and the angle between elbow joint and knee joint is calculated as motion characteristics. Considering the influence of the change of walking speed, the dynamic time warping algorithm is used to calculate the distance of the feature sequence. Finally, the nearest neighbor classifier is used to classify the results accurately. Experimental results show the effectiveness of the proposed method in comparison with other state-of-the-art methods on the public databases CASIA-B and TUM-GAID for gait recognition.
引文
[1]Han J,Bhanu B.Individual recognition using gait energy image[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(2):316-322.
[2]Yang X,Zhou Y,Zhang T.Gait recognition based on dynamic region analysis[J].Signal Processing,2008,88(9):2350-2356.
[3]Lam T H W,Cheung K H,Liu J N K.Gait flow image:a silhouette-based gait representation for human identification[J].Pattern Recognition,2011,44(4):973-987.
[4]Takemura N,Makihara Y,Muramatsu D,et al.Multi-view large population gait dataset and its performance evaluation for cross-view gait recognition[J].IPSJ Transactions on Computer Vision and Applications,2018,10(1):4.
[5]Lee L,Grimson W.Gait analysis for recognition and classification[C]//Proc Fifth IEEE International Conference on Automatic Face and Gesture Recognition,Washington,DC,USA,2002:148-155.
[6]Cunado D,Nixon M S,Carter J N,et al.Automatic extraction and description of human gait models for recognition purposes[J].Computer Vision and Image Understanding,2003,90(1):1-41.
[7]Yoo J H,Nixon M S,Harris C J.Extracting human gait signatures by body segment properties[C]//Proceedings of the Fifth IEEE Southwest Symposium on Image Analysis and Interpretation,2002:35-39.
[8]刘玉栋,苏开娜,马丽.一种基于模型的步态识别方法[J].计算机工程与应用,2005,41(9):88-92.
[9]陈玲.基于视频流的步态识别系统研究与实现[D].广州:暨南大学,2016.
[10]Deng M,Wang C,Cheng F,et al.Fusion of spatialtemporal and kinematic features for gait recognition with deterministic learning[J].Pattern Recognition,2017,67:186-200.
[11]Yang Ke,Dou Yong,Lv Shaohe,et al.Relative distance features for gait recognition with Kinect[J].Journal of Visual Communication and Image Representation,2016,39:209-217.
[12]Wei S E,Ramakrishna V,Kanade T,et al.Convolutional pose machines[C]//Computer Vision and Pattern Recognition(CVPR),2016.
[13]Newell A,Yang K,Deng J.Stacked hourglass networks for human pose estimation[C]//European Conference on Computer Vision,2016:483-499.
[14]施登科.基于人体几何特征的步态识别算法研究及应用平台设计[D].杭州:浙江大学,2017.
[15]Hu M,Wang Y,Zhang Z,et al.Incremental learning for video based gait recognition with lbp flow[J].IEEE Transactions on Cybernetics,2013,43(1):77-89.
[16]Kusakunniran W.Recognizing gaits on spatio-temporal feature domain[J].IEEE Transactions on Information Forensics and Security,2014,9(9):1416-1423.
[17]Yu S,Chen H,Wang Q,et al.Invariant feature extraction for gait recognition using only one uniform model[J].Neurocomputing,2017,239(C):81-93.
