Research on Road Traffic Signal Timing Method Based on Picture Self-learning

  • Guohua Zhu
  • Chi ZhangEmail author
Conference paper
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 127)


Based on the pictures of queuing vehicles at intersections collected by video cameras, a method of traffic signal control based on picture self-learning is proposed. Based on Convolution Neural Network, this paper classifies the queuing length pictures of vehicles with different phase-critical traffic flow, establishes the relationship database between the picture data set and the green light display time of the pictures with different phase-critical traffic flow queuing length categories, and obtains the current phase green light display time of the current cycle on the basis of the relationship database, so as to achieve real-time optimization. The purpose of the signal control scheme. This method does not need the exact traffic flow collected by traffic flow detector, but acquires the pictures of queuing vehicles in different periods and phases, and trains the green light duration to control the traffic in real time.


Intelligent Transportation Systems Traffic signal control Picture self-learning Convolution Neural Network 



The project is supported by Zhangjiang National Independent Innovation Demonstration Zone Special Development Foundation (Number 201705-JD-C1085-072).


  1. 1.
    Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep convolutional neural networks. In: International Conference on Neural Information Processing Systems, pp. 1097–1105. Curran Associates Inc. (2012)Google Scholar
  2. 2.
    Wang, X., Gao, Y., Liu, D.: Design of intelligent traffic light control system based on machine vision. Autom. Appl. 11, 38–39 (2015)Google Scholar
  3. 3.
    Tang, Y.: A traffic signal control method based on deep Convolution Neural Network. China Patent: CN106023605A, 12 October 2016Google Scholar
  4. 4.
    Sermanet, P., Chintala, S., LeCun, Y.: Convolutional Neural Networks applied to house numbers digit classification. In: 2012 21st International Conference on Pattern Recognition (ICPR), pp. 3288–3291. IEEE (2012)Google Scholar
  5. 5.
    Sermanet, P., LeCun, Y.: Traffic sign recognition with multi-scale convolutional networks. In: The 2011 International Joint Conference on Neural Networks (IJCNN), pp. 2809–2813. IEEE (2011)Google Scholar
  6. 6.
    OpenITS (EB/OL) (2018).

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  1. 1.Ze Yi Traffic Engineering Consulting (Shanghai) Co., Ltd.ShanghaiChina
  2. 2.Yan Yun Network Technology (Shanghai) Co., Ltd.ShanghaiChina

Personalised recommendations