Research Of License Plate Recognition System Based On Certain Digital Image Processing Technology

License plate recognition system is widely used in car parking management, highway speed detection system,traffic management system, vehicle monitoring system. Therefore, the license plate recognition system can effectively serve the traffic management system to improve traffic management efficiency.License plate recognition system includes license plate location, license plate character segmentation; character recognition. Considering license plate recognition system for domestic and international research status, the paper gives a comprehensive introduction and an in-depth study about license plate location algorithm and character segmentation and recognition algorithms.Before the license plate location, RGB image is translated into gray image. Weather and light have a greater impact on the RGB image, so accuracy is lower when we use color features to locate the plate; Moreover, color images require more storage space, which will make the processing speed slow. Therefore, in the preprocessing; RGB images collected are translated into gray image.License plate location is the key technology in license plate recognition system. License plate location accuracy directly influences the license plate identification effect. License plate location goal is to make the license plate area segmented out from the vehicle's image. In the license plate location part, the paper put forward a location algorithm based on the texture characteristics and color characteristics of the license plate. The algorithm firstly makes use of license plate texture feature to locate license plate roughly, including the edge detection, mathematics morphological processing, etc. Finally, using color characteristics of vehicle license plate area for accurate positioning, mainly including space transform, boundary determination. 80 images were tested and the results show that the method is accurate and fast.In character segmentation part, according to the characteristics of license plate, we use an improved level projection method for character segmentation. First, from head to the next line, when the first black pixel is detected, we will think that its row is the border place of the character. And again from upper upwards scanning, when the scanning to the first black pixel, we will do as character in the lower; Secondly, using of the horizontal projection method for a coarse division which is about the minimum of cutting points. Finally, the paper adopts a horizontal projection method for a single character image, scanning images of characters from left to right in horizontally. When a row of black lace which pixel sum is smaller than a certain threshold, which detects the border of the character. This method to overcome character adhesion and characters-disconnected problem is very effective.In character recognition, the paper studies the part about contour matching method based on the statistical characteristics. Identification tests act on the last character of license plate. First we calculated respectively distance from the first black pixel to the corresponding boundary in the four directions of image, which is thought as the contour features. Secondly, we will take outline feature as the input value of a discriminate function, and use discriminate function to identify characters.