License Plate Character Segmentation Algorithm Based on Improved Regression Model

Abstract

Abstract Many character segmentation algorithms in license plate recognition system are not very good for segmentation of complex characters, and there is a phenomenon of character sticking, which results in slow recognition speed and not very good recognition effect of the license plate with complex characters. For such problems, a segmentation algorithm based on the improved regression model on statistical characteristics is presented. Compared with other algorithms, experimental results show that less time is consumed and segmentation effect is better. The recognition rate of license plate can be improved effectively.