Research And Implementation Of License Plate Recognition Technology

Automatic license plate recognition system is a specific computer vision system based on the specific goals of the vehicle license. It is one of the important research topics in the fields of computer vision and pattern recognition technology in intelligent transportation applications, and it is the important part in intelligent traffic management as well.License Plate Recognition System includes the license plate images acquisition, preprocessing, character recognition and extraction of characters. This dissertation studies the critical technologies of automatic plate recognition, including plate image preprocessing, plate location, character segmentation and character recognition. The author completed the experiment and has made some improvement on some algorithms.Image preprocessing is the most critical step in license plates and identification. Based on the integrated treatment of the various pictures, This dissertation sums up the pretreatment methods, such as gray, filtering, median filtering and histogram correction, histogram equalization, it presents an improved filtering method, and eliminated the noise of images nicely;In the part of the plate location, based on the study of plate features and location positioning technology, this dissertation has completed the plate location methods for detecting edge and mathematical morphology with a real number plates and has extracted plate regions accurately;Before the plate characters segmentation, the dissertation proposed an improved Bernsen algorithm of license plate binary and tilt correction. By combining edge detection and projection , the improved Bernsen algorithm uses Prewitt template to correct tilt, this method is simple, fast, denoise robustness. In the part of plate segmentation, it has completed the characters segmentation by combination of regional growth and projection strategy. In the part of character recognition, based on pattern matching, the dissertation proposes an improved algorithm, which is simple, but has strong resistance to noise and higher recognition rate.Finally,the work of the dissertation is summarized and the further work is proposed.