The Open Automation and Control Systems Journal

2014, 6 : 349-354
Published online 2014 December 26. DOI: 10.2174/1874444301406010349
Publisher ID: TOAUTOCJ-6-349

Robust Vehicle Detection Based on Cascade Classifier in Traffic Surveillance System

Y. Tang , Y. C. Xu and C. Z. Zhang
School of Automobile and Traffic Engineering, Nanjing Forestry University, Nanjing, 210037, P.R. China. Nanjing, 201137, P.R. China.

ABSTRACT

Vehicle detection based on static images are highly practical and directly applicable for vehicle feature extraction and recognition in a traffic surveillance system. This paper will introduce the processing of automatic vehicle detection based on machine learning algorithm. Firstly, Haar-like feature is used to represent the appearance of vehicle, and then a learning algorithm, based on AdaBoost, is to train the strong classifier, at last a method for combining strong classifiers in a cascade is proposed, which allows background regions of the image can be removed quickly. The experimental result shows that the our classifier can achieve good performance of vehicle detection, its detection rate is more than 97% and its false alarm rate is only 3.4%.

Keywords:

AdaBoost algorithm, Cascade, Haar-like feature, Intelligent transportation vehicle detection.