License plate detection and recognition in complex scenes using mathematical morphology and support vector machines

In this paper we propose a highly reliable license plate detection and recognition approach using mathematical morphology and support vector machines (SVM). The approach is composed of three main stages including license plate detection, character segmentation and recognition. A preprocessing step is applied to improve the performance of license plate localization and character segmentation in case of severe imaging conditions. The first and second stages utilize edge detection, mathematical morphology followed by connected component analysis. While SVM is employed in the last stage to construct a classifier to categorize the input numbers of the license plate into one of 9 classes. The algorithm has been applied on 208 car images with different backgrounds, license plate angles, distances, lightning conditions, and colors. The average accuracy of the license plate localization is 97.60%, 90.74% for license plate identification, and 97.89% for number recognition.