Research On Vehicle License Plate Location And Character Segmentation Under Complicated Background

With the rapid development of transportation in our country, the ITS (Intelligent Traffic Management System) becomes a hotspot problem which is being paid high attention. License plate recognition is a key component of the ITS. It concludes license plate location, character segmentation and character recognition. The license plate location and character segmentation play a crucial roles in the character recognition. Although this technology has been researched for a period of times, there are still some problems in existing algorithms, such as the algorithm's environmental adaptability is bad and the recognition accuracy and the processing speed are low in a complex background. So the Location and Segmentation of the license plates is always the hot research topic of this domain.After analyzing the popular methods; this paper is mostly studying license plate location and character segmentation that could be used under complex background. The mainly works that have been done in this paper include:A new vehicle license plate location algorithm based on mufti-characteristics is introduced at the license plate location phase. This algorithm makes full use of the color, texture, geometric characteristics and locates the license plate by six steps: edge detector, binary quantifying, filtering, areas closing, areas analysis, color matching. In the process of filtering, this paper designs a multi-scale model filter. After processing, the noise is reduced effectively. The experimental result shows that this algorithm has a fast, efficient performance of locating vehicle license plate under the complex background.According to the rules of the characters of a vehicle license plate and the geometrical features of the characters, an approach for character segmentation of vehicle license plates is proposed which is based on the location and unify in level direction of the second and third character. At last, this algorithm combines three tradition methods together by Maximum variance betweens clusters to locate the license plate.