Research On License Plate Location Algorithm Of License Plate Recognition System

With the rapid development of the economy, transportation enter into the rapiddevelopment stage, growth in road transport and environmental resources havebecome increasingly prominent, intelligent transportation systems came into being.License Plate Recognition System as an important part of the intelligent transportationsystem, which mainly consists of four parts of the image acquisition, license platelocation, character segmentation, character recognition.In this paper, analyzed of the license plate features, implemented two locationalgorithm based on gray-scale image and summarizes the advantages anddisadvantages. Then proposed license plate location algorithm which integrated colorcharacteristics. The algorithm overcome the limitations of a single feature location,and main contents are as follows:1. Analyzed and summarized the characteristics of the license plate: geometry,gradation characteristics, structural features, color features, border features, analyzesthe diversity of the license plate and difficulties.2. Studied and implemented two location algorithm based on gray image, whichis based localization algorithm scan lines and localization algorithm based onmorphology, the former combined geometric and gradation characteristics of thelicense plate, the latter combined license geometry and structural features. Analyzedof experimental results, when the picture had uneven distribution of gray or existedsimilar characteristics with license, then positioned failure, at this point the charactercolor can distinguish interference area. So made full use of the color characteristicsand combined with other features to locate license plate, the license plate locationalgorithm based on character edge color.3. The algorithm combined the features of the license plate color, geometry,gradation characteristics, structural features, the research focused on extractinglicense plate color features. chose HSV color space model, counted color threshold ofbackground and character of license plate, then extracted character edge color byusing color image binarization principle and logic operation. Combined with thelicense plate gradation characteristics, structural features, geometric features tocomplete regional location upon completion of the character edge extraction.4. Showed and analyzed the experimental results based on the specialcircumstances picture, such as when the body color was consistent with background color of license plate, when license was defaced, when license plate was faded. Tookmany actual pictures of a unit, combined with experimental data, analyzed andcompared with the three algorithms, summarized the advantages and disadvantages,then verified the suitability of the algorithm...