Design based on improved license plate location and projection character segmentation system

Aiming at the problems such as low license plate positioning rate and poor character segmentation effect based on traditional methods, focuses on a precise license plate positioning method based on color characteristics. It makes full use of the visual perception ability of human eyes, filters out the blue pixels in the license plate through color model conversion and setting a reasonable range of H, S and V components, so as to achieve accurate positioning. The localization rate of this algorithm is 96.7%. In this paper, a new projection character segmentation process is proposed, which combines morphological connected domain idea with projection method. The above method has the advantages of good accuracy and robustness. Compared with the general algorithm, the positioning rate increases by 11.5%, improves the segmentation effect, and can meet the engineering requirements of license plate recognition system.