摘要
提出一种针对车牌识别算法要求的车牌图像质量综合评价方法。其中根据车牌识别算法的普遍要求,归纳出了四种车牌图像质量评价因子:亮度偏移、亮度分布一致性、对比度、清晰度。最后使用主成份分析方法将四种评价因子综合起来建立得到对车牌图像的总体质量评价标准。采用这个评价标准,可以在无参考的情况下对单张车牌图像给出质量评价,为降低系统误差提供量化的依据,从而到达为识别算法提供高质量的输入图像的目的。
A new quality assessment method of license plate image has been presented in this thesis. Four image quality factors have been brought out ： brightness deviation, distribution consistency of brightness, contrast and definition. And then principal component analysis （PCA） has been used to confuse the four quality factors to obtain the comprehensive evaluation of the license plate image quality. This quality assessment method is no-reference and can be used to assess quality of a single license plate image. This method can be used to improve the system status and get high quality image for license plate recognition.
出处
《科学技术与工程》
北大核心
2012年第33期8915-8918,8932,共5页
Science Technology and Engineering
基金
中央高校基本科研业务费项目(2010SCU11052)资助
关键词
智能交通系统
车牌识别
车牌图像质量
无参考图像质量评价
主成份分析
intelligent transportation system
license plate recognition
license plate image quality
no-reference image quality assessment
PCA
