一种基于SVM的真伪车牌分类算法

An Algorithm Based on SVM to Classify True or False License Plates

车牌定位算法往往会定位出一些伪车牌候选区域,为准确区分出车牌候选区域中的真伪车牌,提出了一种基于支持向量机(SVM)的真伪车牌分类算法。该算法首先提取出车牌候选区域图像的纹理特征和几何特征,构成特征向量集;然后根据特征向量训练SVM分类器;最后利用训练好的分类器来实现对真伪候选车牌区域图像进行分类判断,得到真车牌区域。实验表明,算法对车牌候选区域分类准确率高达99.3%,且具有较强的抗干扰能力和鲁棒性。

The license plate location algorithm always locates some pseudo candidate regions of license plates. To distinguish the true or false license plates accurately from candidate regions of license plates,in this paper,we propose a classification algorithm that is based on support vector machine（ SVM）. Firstly,the algorithm extracts license plate image texture and geometrical features of candidate regions to form an eigenvector set. Next,it trains the SVM classifier. Finally,it classifies the license plates＇ candidate regions via the well trained classifier. Experimental results show that the algorithm can classify the license plates＇ candidate regions with accuracy up to 99. 3%,and has a strong anti-interference ability and robustness.