Automatic Traffic State Recognition from Videos Based on Autoencoder and <i>k</i>-means Clustering

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (1574 KB) HTML ()
输出: BibTeX | EndNote (RIS)

摘要 An automatic video traffic state recognition method based on an autoencoder and k-means clustering is proposed to timely and effectively recognize road traffic state. First, candidate autoencoders are established through a reasonable optimization of the structural parameters on the input data dimensions, number of hidden layers, and dimensions of dimension-reduced data through cross-examination. Then, three image data sets are formed with 1500-4500 sample images. On this basis, the candidate autoencoders are trained and tested; thus, the best autoencoder AE^* is proposed according to precision, recall, and F1-value. Lastly, four traffic state recognition models are constructed by combining AE^* with k-means clustering, Support Vector Machine (SVM), Linear Classifier, and DNN Linear Classifier (Deep Neural Network with Linear Classifier), which are named as AE^*-kmeans, AE^*-SVM, AE^*-Linear, and AE^*-DNN_Linear, respectively. The models are trained and tested on the basis of the three image data sets. Results show that the four models' average precision in terms of precision and recall is 91.9%-92.7%, and their average recall is 91.6%-92.6%, while AE^*-kmeans performs best or second to best in terms of precision and recall. With regard to the comprehensive evaluation index F1-value, AE^*-kmeans achieves 92.4%, a little lower than 92.7% of AE^*-SVM, and better than AE^*-DNN_Linear (92.1%) and AE^*-Linear (91.8%). Given that k-means is an unsupervised clustering method, compared with AE^*-SVM, AE^*-Linear, and AE^*-DNN_Linear, AE^*-kmeans can reduce the workload, such as manual data calibration and supervised training and cut down calculation cost. Meanwhile, AE^*-kmeans also obtains a good traffic state recognition result. Therefore, this mechanism has high practical significance for an accurate real-time extraction of a video traffic status.

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	张媛媛
	王玉婷
	彭博
	唐聚
	谢济铭

关键词 ： ITS, video traffic state recognition, deep learning, urban road, autoencoder

Abstract：An automatic video traffic state recognition method based on an autoencoder and k-means clustering is proposed to timely and effectively recognize road traffic state. First, candidate autoencoders are established through a reasonable optimization of the structural parameters on the input data dimensions, number of hidden layers, and dimensions of dimension-reduced data through cross-examination. Then, three image data sets are formed with 1500-4500 sample images. On this basis, the candidate autoencoders are trained and tested; thus, the best autoencoder AE^* is proposed according to precision, recall, and F1-value. Lastly, four traffic state recognition models are constructed by combining AE^* with k-means clustering, Support Vector Machine (SVM), Linear Classifier, and DNN Linear Classifier (Deep Neural Network with Linear Classifier), which are named as AE^*-kmeans, AE^*-SVM, AE^*-Linear, and AE^*-DNN_Linear, respectively. The models are trained and tested on the basis of the three image data sets. Results show that the four models' average precision in terms of precision and recall is 91.9%-92.7%, and their average recall is 91.6%-92.6%, while AE^*-kmeans performs best or second to best in terms of precision and recall. With regard to the comprehensive evaluation index F1-value, AE^*-kmeans achieves 92.4%, a little lower than 92.7% of AE^*-SVM, and better than AE^*-DNN_Linear (92.1%) and AE^*-Linear (91.8%). Given that k-means is an unsupervised clustering method, compared with AE^*-SVM, AE^*-Linear, and AE^*-DNN_Linear, AE^*-kmeans can reduce the workload, such as manual data calibration and supervised training and cut down calculation cost. Meanwhile, AE^*-kmeans also obtains a good traffic state recognition result. Therefore, this mechanism has high practical significance for an accurate real-time extraction of a video traffic status.

Key words： ITS video traffic state recognition deep learning urban road autoencoder

收稿日期: 2019-12-24

基金资助:Supported by the National Natural Science Foundation of China(No.61703064); Chongqing Research Program of Basic Research and Frontier Technology Innovation(Nos.cstc2017jcyjAX0473, cstc2018jscx-msybX0295); Scientific Research Project of Key Laboratory of Urban ITS Technology Optimization, Traffic System & Safety in Mountain Cities(Nos. 2017KFKT01, 2018TSSMC05)

通讯作者: ZHANG Yuan-yuan E-mail: 1605968020@qq.com

引用本文:

张媛媛, 王玉婷, 彭博, 唐聚, 谢济铭. Automatic Traffic State Recognition from Videos Based on Autoencoder and k-means Clustering[J]. Journal of Highway and Transportation Research and Development, 2020, 14(4): 81-88.
ZHANG Yuan-yuan, WANG Yu-ting, PENG Bo, TANG Ju, XIE Ji-ming. Automatic Traffic State Recognition from Videos Based on Autoencoder and k-means Clustering. Journal of Highway and Transportation Research and Development, 2020, 14(4): 81-88.

链接本文:

http://journal10.magtechjournal.com/Jwk_gljtkj_en/CN/ 或 http://journal10.magtechjournal.com/Jwk_gljtkj_en/CN/Y2020/V14/I4/81

[1] WAN Fang, NIN Shu-yun, ZHU Li-li, et al. Real-time Traffic State Expression based on DSRC and Point Detector Data Fusion[J]. Journal of Highway and Transportation Research and Development, 2017, 34(8):123-129. (in Chinese)
[2] PI Xiao-liang, WANG Zheng, HAN Hao, et al. Study on Application of Traffic State Classification Method based on Information Collected by Loop Detector[J]. Highway Traffic Technology, 2006, 23(4):33-38.
[3] LI X, SHE Y, LUO D, et al. A Traffic State Detection Tool for Freeway Video Surveillance System[J]. Procedia-Social and Behavioral Sciences, 2013, 96:2453-2461.
[4] BYSHOV N, SIMDIANKIN A, USPENSKY I, et al. Method of Traffic Safety Enhancement with use of RFID Technologies and its Implementation[J]. Transportation Research Procedia, 2017, 20(20):107-111.
[5] KONG X, XU Z, SHEN G, et al. Urban Traffic Congestion Estimation and Prediction based on Floating Car Trajectory Data[J]. Future Generation Computer Systems, 2015, 61(C):97-107.
[6] MORTON O'Kelly, MATISZIW T, LI R, et al. Identifying Truck Correspondence in Multi-frame Imagery[J]. Transportation Research Part C Emerging Technologies, 2005, 13(1):1-17.
[7] REINARTZ P, LACHAISE M, SCHMEER E, et al. Traffic Monitoring with Serial Images from Airborne Cameras[J]. ISPRS Journal of Photogrammetry & Remote Sensing, 2006, 61(3/4):149-158.
[8] ANGéLICA L, MANFREDI G, NIEDDU L. An Algorithm for the Recognition of Levels of Congestion in Road Traffic Problems[J]. Mathematics and Computers in Simulation, 2009, 79(6):1926-1934.
[9] LIAO L C, JIANG X H, ZOU F M, et al. An Automatic Recognition Approach for Traffic Congestion States based on Traffic Video[J]. Journal of Highway and Transportation Research and Development, 2014, 8(2):72-80.
[10] QUIROS A R F, BEDRUZ R A, UY A C, et al. Machine Vision of Traffic State Estimation using Fuzzy Logic[C]//IEEE Region 10 Conference, Marina Bay Sands, Singapore, 2016:2104-2109.
[11] CUI Hua, YUAN Chao, WEI Ze-fa, et al. Traffic State Recognition using State Images and FCM[J]. Journal of Xidian University, 2017(06):85-90. (in Chinese)
[12] WANG J, HE H, PROKHOROV D V. A Folded Neural Network Autoencoder for Dimensionality Reduction[J]. Procedia Computer Science, 2012, 13:120-127.
[13] GAO S, ZHANG Y, JIA K, et al. Single Sample Face Recognition via Learning Deep Supervised Autoencoders[J]. IEEE Transactions on Information Forensics and Security, 2015, 10(10):2108-2118.
[14] ZHAO C, WAN X, ZHAO G, et al. Spectral-spatial Classification of Hyperspectral Imagery based on Stacked Sparse Autoencoder and Random Forest[J]. European Journal of Remote Sensing, 2017, 50(1):47-63.
[15] GE P, REN C X, DAI D Q, et al. Dual Adversarial Autoencoders for Clustering[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020,31(4):1417-1424.
[16] KINGMA D P, BA J. Adam:A Method for Stochastic Optimization[J]. Computer Science, 2014.
[17] SUN Z, LI Z, ZHAO Y. Traffic Congestion Forecasting based on Possibility Theory[J]. International Journal of Intelligent Transportation Systems Research, 2016, 14(2):85-91.
[18] CHENG H, KOC L, HARMSEN J, et al. Wide & Deep Learning for Recommender Systems[C]//Conference on Recommender Systems, New York, America, 2016:7-10.