Real-time license plate detection under various conditions

Abstract

This paper proposes an algorithm for real-time license plate detection. In this algorithm, the relatively easy car plate features are adopted including the simple statistical feature and Harr-like feature. The simplicity of the object features used is very helpful to real-time processing. The classifiers based on statistical features decrease the complexity of the system. They are followed by the classifiers based on Haar-like features, which makes the final classifier invariant to the brightness, color, size and position of license plates. The experimental results obtained by the proposed algorithm exhibit the encouraging performance. © Springer-Verlag Berlin Heidelberg 2006.