Study On The License Plate Localization And Character Segmentation Based On Color Characteristics

In the field of intelligent transport，for the requirement of vehicle managementsuch as standardization,intellectualization and so on， license plate recognitionsystem(LPR)has become one of the important research subjects of computer vision andartificial intelligence,and has broad application prospects.License plate recognition system mainly consists of three major modules,respectively is image acquisition and pretreatment, license plate location andsegmentation, character recognition. This paper focuses on license plate location andsegmentation, presents a fast and reliable solutions that lay a foundation for subsequentaccurate recognition. In order to improve image quality, making the subsequent accuratelicense plate segmentation easier, this paper first studies the image pretreatmenttechnique such as gray image equalization and noise processing which has grateinfluence on the image processing. Secondly, this paper has analyzed the existinglicense plate location technology, according to the character texture, shape and colorcharacteristics of domestic license plate, presents a method for coarse positioninglicense plate based on HSV color space and color texture. And then according to theprior knowledge of license plate, we can use the method of color pair’s validation toconfirm the license plate region as to realize the quick location of the license plate.Finally, after the realization of license plate location, the paper puts forward a licenseplate character segmentation method based on projection feature, effectively solves theproblem of extraction of license plate characters. Based on the above steps, on the realexperimental test of vehicle images, experimental results show that the proposed licenseplate location method is fast, higher accuracy and is an efficient license plate locationalgorithm.