【摘要】 智能车牌定位是一个经典而又具有重大使用价值的课题。针对现实生活中车牌的纹理及颜色特征,提出综合利用纹理信息及颜色信息实现车牌的快速定位。先充分利用纹理特征,结合隔列处理思想和类投影技巧,在将原始图像连续的纹理信息空间转换成离散的短线条信息空间后,统计预处理过程中的短线密度,取密度最大区域,即为车牌所在区域,实现初步定位,再在初定位的区域结合车牌底色先验知识,延展初定位区域,进行颜色匹配,进而可以实现精确车牌定位,方法解决了仅仅依靠颜色信息或纹理信息车牌定位准确率低的问题。算法定位速度快,鲁棒性性强,准确率高,在现实中有很强的实用性。更多还原

【Abstract】 Car license plate location is not only a classical matter of image segmentation but also a subject with great practical value.In view of the car license plate characteristic in the real life,this paper has proposed a method which utilizes both texture and color information to implement fast localization of the car license plate.Primary location was implemented using texture characteristic combined with the concept of separated row processing and skills of similar projection.Then accurate location was implemented through color matching with the pre-acquired color information in primarily located region.This method is better than others which use only color or texture information and is proved effective with high speed and accuracy.更多还原