摘要
车牌定位是车牌识别系统的关键技术,车牌定位的准确与否直接影响车牌识别的结果。车牌区域的颜色特征是车牌的重要信息,针对车牌区域的颜色特征,论文提出了一套利用车牌区域的伴生与互补特征进行定位的算法。该算法利用车牌区域的颜色特征,快速定位到与车牌颜色有关的区域;然后利用车牌区域伴生与互补特性快速去除具有与车牌区域相同颜色的其他非车牌区域;最后使用投影积分进行车牌的精确定位。通过对200幅从交通卡口获取的真实的彩色图像进行试验,准确定位率为98%。
The vehicle license plate location technique plays an important role in the vehicle license plate recognition system;the accuracy will influence the plate recognition result.The color feature is the important information of the license plate,for this character the paper shows a method to locate the license plate using the concomitant and complementary color feature in the license plate region.First,to locate all color regions similarly with the license plate.Second,to remove body regions with similar color using concomitant and complementary color feature.At last,to locate the license plate accurately by its integral projection figures.The experiments on 200 automobile realistic RGB color images that were taken under various conditions from traffic stops show the accuracy of 98%.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第1期206-208,211,共4页
Computer Engineering and Applications
基金
国家自然科学基金资助项目(60541001)
全国优秀博士学位论文作者专项基金项目(200443)。
关键词
车牌定位
伴生
互补
颜色特征
vehicle license location
concomitant
complement
color feature
