摘要
通过对车牌纹理特征和灰度特征的研究,提出基于车牌字符垂直笔划检测的自适应反色判断算法和基于分块平移的二值化算法,解决目前车牌反色判断适应性不够和对光照不均、对比度低等车牌的快速二值化问题。对自然环境中采集得到的3653例车牌图像数据库进行实验,实验结果表明,反色判断算法的准确率为99.48%,分块二值化方法对阴影、低对比度等干扰具有较好的适应性,两个算法都满足实时性要求,可以应用到实际车牌识别系统中。
To deal with the problem of insufficient adaptiveness in judging reverse color and binaryzation in licence plate with nonuniform illumination and low contrast,an adaptive method to judge whether the plate needed reverse color based on the detection of vertical stroke and a method of image binaryzation based on the image block partition and shift were proposed by studying texture feature and gray feature in license plate.The algorithms were tested with 3653 images of license plate captured under the natural condition,and the results show that an accurate rate of 99.48%is obtained using the method of judgment of reverse co-lor.The binaryzation method based on block partition has good adaptation to the image with shadow and low contrast.Both methods meet the requirement of real-time and can be applied to real-time LPR system.
出处
《计算机工程与设计》
北大核心
2016年第2期534-539,共6页
Computer Engineering and Design
基金
广西研究生教育创新计划基金项目(YCS22013043)
广西多源信息挖掘与安全重点实验室基金项目(13-A-03-02)
关键词
反色判断
字符笔划检测
车牌二值化
分块二值化
车牌识别
reverse color judgement
character stroke detection
binaryzation of license plate
binaryzation of block partition
license plate recognition
