Chinese-style Plate Recognition Based on Artificial Neural Network and Statistics

Abstract

In this paper we propose an algorithm for the recognition of Chinese-style vehicle license plates. Because the resolution of videos is much lower than that of static images, the approaches designed for recognizing license plates in the static picture don’t yield satisfying results when processing videos. In addition, the traditional approaches only take a single snapshot per plate to recognize, which is insufficient to yield accurate recognition results for that plate. In order to solve this problem, our approach processes multiple frames and generates a series of results per plate. And it combines artificial neural network, which identifies characters in the snapshot, and a simple statistics which finds the most appealing results among the result sequences. The statistics analysis can confirm the accuracy of identification. Finally, this approach has been proved to be accurate and efficient by the experiment.