基于颜色特征的真假车牌识别方法研究

摘要

The paper analyzes theoretically the color differences of true and fake licenses to propose a method that can distinguish true license and fake license by the SVM technology. This method can extract the license image to convert it to RGB and HSI components. Aiming at R and S components that have obvious differences between true and fake licenses , this method can make the hierarchical processing in quantity to get the data. Then, it takes the data, the mean and variance of R components as the eigenvectors to design SVM classifier for the purpose of identifying and classifying true and fake licenses. The experiments show this method has certain effects on identifying true and fake licenses , it can provide a new thought for the identification of true and fake licenses.%从理论上分析了真、假车牌在颜色特征上的差异,提出了一种通过SVM技术辨别真车牌和假车牌的方法.该方法将提取到的车牌图像转换为RGB分量和HSI分量,对真、假车牌间存在明显差异的R分量和S分量作分级量化处理,将得到的数据和R分量均值、 方差作为特征向量设计SVM分类器,对真、假车牌进行识别与分类.通过试验表明该方法对真、假车牌有一定的鉴别效果,为真、假车牌的识别提供了新思路.