Skip to Main Content
License plate character segmentation is a crucial process of license plate recognition. During the process of image acquisition, because of light condition and the change of shooting distance and the diversity of the plate's format, the task become very challenging. Therefore, many algorithms are proposed under certain conditions, such as a fixed shooting distance or no illumination change. The purpose of this paper is to classify and assess the technique which has been developed for license plate character segmentation. Execution time, accuracy of segmentation, the robustness of algorithm will be reported. At last, the technique based on the connected component correction is proposed in the thesis. The result manifests that the accuracy of character segmentation could reach 84.4%.