Font Size: a A A

Complex License Plate Recognition System Based On Shape Context

Posted on:2012-11-14Degree:MasterType:Thesis
Country:ChinaCandidate:Y P GuoFull Text:PDF
GTID:2248330395962336Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
License Plate Recognition is one of the most important technologies in intelligent transportation. It mainly includes three parts, which are plate localization, plate segmentation and character recognition. This work deals with three fields of image processing, machine vision and neuron network. This problematic includes mathematical principles and improved algorithms, which ensure the accuracy of plate localization, plate segmentation, character normalization and recognition.The main parts of the paper can be divided into several parts:(1) Introduction to the background and realistic meaning of the license plate recognition. Brief explain of present research on license plate recognition at home and abroad. Analysis of the license plate particularity and the develop prospects of our country,(2) According to the prior knowledge of license plate of our country, we can get factors that affect how to choose the best plate candidate. The plate candidates are extracted based on improved projection algorithm.(3) The paper proposes an iterative segmentation algorithm, which makes full use of the prior knowledge of the plate, it can achieve an excellent segmentation result even if the snapshot has complicated background, or it is captured in various light conditions or different scales.(4) The contour of object is prominent and stable, even if it is under complicated environment. Therefore, the feature of the segmented characters is extracted based on algorithm of shape context, which is invariant towards different scales and rotation. After the character feature is extracted, we can use shape match method or neuron network to implement the recognition process.(5) According to country-specific rules for the plate, we can evaluate the validity of that plate recognition towards based on these rules. Different classes of characters are required at different locations, so we can correct some failed recognition result by syntactical analysis of the recognized plate. There is an experiment that shows the recognition result of our system, it also gives a definition to evaluate how successfully the licensed plate is recognized. The experiment demonstrates that, the whole process of recognizing plate is very useful in practical, it can be widely used.(6) Summarize the pros and cons of the license plate system of this paper, point out the future research direction and problems should be paid attention to.
Keywords/Search Tags:projection method, plate localization, iterative segmentation, shape context, syntacticalanalysis of recognized plate
PDF Full Text Request
Related items