License Plate Recognition Technology Complex Background

The importance of license plate recognition technology in intelligenttransportation field has increased a lot in recent years. License plate recognitiontechnology can be used in the district parking management system, electroniceye system in the important transportation hub, highway management systemand etc. It brings a lot of convenience and protection for public safety andnational development. With the increasing complexity of traffic environment,the current license plate recognition products cannot satisfy the user’s real-timerequirement. So how to obtain the balance between accuracy and speed oflicense plate recognition in complex background is a hot topic in recent years,there are still many problems remain to be further explored and improved.This paper discusses the key algorithm in license plate recognitiontechnology under complicated background. The work I did include:1)This paper introduced the commonly used license plate imagepreprocessing technology, analyzes the advantages and disadvantages of theexisting license plate localization algorithm. According to the difficulties undercomplex environment such as the proportion of the license plate area in thewhole image is small, and many texture features, color features in noise areasimilar to the regional features of the plate area in the image, it proposed a fastand accurate license plate localization algorithm based on edge detection,morphological filter and the license plate color features.2) This paper analyzes the characteristics of license plate characters inChina and compares the common license plate recognition charactersegmentation algorithm. Aiming at the problem of license plate characterssegmentation under complicated background, then presents a license plate tiltcorrection algorithm based on phase marshalling method and fast probabilisticHough transform, on this basis, proposes the connected component analysis anda two-way combination of one-third of the projection of license plate charactersegmentation algorithm to achieve the fast and accurate character segmentation of licenses plate.3) On the basis of summarizing the strengths and weaknesses of the currentcharacter recognition algorithm, introduces a invariable and robust shapedescription factor, shape context, proposes a character recognition algorithmwith an improved pattern matching on shape context, which achieves higherlicense plate character recognition rate and recognition speed.The simulation results prove the effectiveness of the algorithm undercomplicated background. It can be used in the license plate recognition system.