摘要
针对夜间的车牌识别,提出了一种基于车灯定位的识别多个车牌的流程方法。该方法首先利用车灯坐标来对图像中多个车牌区域进行定位,进而对局部区域利用边缘检测,形态学技术,使车牌成唯一存在多次跳变的区域,车牌定位之后又利用改进的模板匹配的算法,提高了误判率,从而提供了一套夜间多车牌识别的方法。试验结果表明,两辆车的车牌识别的准确率在95%以上,并能适应于夜间的不同环境。
In order to solve the license plate recognition at night,a method is proposed based on headlight position to recognize multiple plates.At first we use car lamp coordinates to locate the multiple plate regions in images.Then process the local region which contain plates and use edge detection,morphology technology to make the plate become the only one region which exist many time jumps.After license plate location we use modified template matching algorithm to improve the false reject rate.From the above,we provide a procedure of night license plate identification.The test result shows that the two car license plate recognition accuracy is above 95%,and can adapt to different environment.
出处
《微计算机信息》
2011年第1期278-280,共3页
Control & Automation
关键词
夜间背景
车灯定位
边缘检测
数学形态学
模式匹配
night conditions
headlight position
edge detection
mathematical morphology
template matching
