The Research Of Vehicle License Plate Recognition Based On Color Division

With the development and progress of the economic and technology, the transportation has become more and more important in the human’s daily life. Intelligent transportation system has become an inevitable trend in the new age because of it’s widely applications.A new image processing algorithm based on color division was proposed in this paper. The main idea of color division is changing all the pixels in plate area into standard color and ignoring all the other colors not in license plate through some transformation of all the image pixels. There are totally6colors in the image including5colors in license plate and1other color after color division.In the stage of license plate location, a new locating method based on edge color pairs was proposed. For any kind of vehicle license plate, it always has a fixed kind of edge color pairs which are composed of license plate background color and characters color. So it is necessary to search color pairs in level and vertical direction. The license plate would be accurately located by coarse location and accurate locationProjection was used to separate all the plate characters. We can get a chart with peaks and valleys of image by using projection. Through combining the ration of character width and height, all the characters could be separated. After character separating, template matching was used to recognize all the characters.Experiments show that there would be less interference in vehicle image after color division, especially the interference from environment lightening. It is good for the next stages. Coarse locating and accurate locating were successfully used in license plate locating. It has good performance and combining some priori knowledge of license plate can locate the license plate area successfully and accurately. In character separation, some priori knowledge of character was used to help separating, all algorithms have good performances.