基于粒子群优化的车牌识别算法研究

Study on License Plate Recognition Approach Based on Particle Swarm Optimization

针对环境噪声严重影响车牌的识别问题,基于字符特征向量和粒子群优化设计一种车牌识别算法。借助垂直投影法、自适应阈值方案、方向梯度直方图法等,对车牌字符进行分割和提取字符的特征向量。依据字符特征向量样本和支持向量机建立字符识别准确率模型,并基于粒子群优化算法建立求解该模型的车牌识别算法。比较性的数值实验显示,该算法能有效提升车牌识别的准确率,且字符特征向量对车牌识别有极大影响。

This work,based on particle swarm optimization designs a license plate recognition algorithm.License plates’character segmentation and feature extraction are executed in terms of the methods of vertical projection,adaptive threshold,directional gradient histogram and so forth.A character recognition accuracy model is developed by means of a character feature vector sample and support vector machine,solved by a novel license plate recognition algorithm.Numerically comparative experiments show that the approach can effectively enhance the accuracy of license plate recognition while license plate recognition’effect depends greatly on character features.