基于三次定位过程的车牌精确定位算法设计

摘要

为了实现对车牌区域的精确定位,提出了一种基于三次定位过程的车牌定位算法,综合改进了数学形态学和纹理分析等方法.对自然环境下摄像头采集的车辆图像进行预处理后,使用像素统计法对车牌进行粗略定位,使用纹理分析法去除车牌上下边框和干扰,使用阈值法去除车牌左右边框和干扰,得到精确的车牌区域.另外还对图像的预处理进行简化,保证实时性要求.实验结果表明,改进后的算法能够对车牌进行实时、精确定位,满足了系统的要求.%To realize accurate location of license plate,a method based on three-step location is presented.The mathematical morphology and texture analysis method are integrated and improved.Firstly,images of vehicles are pretreated under the natural environment,and rough location of the plate with pixel statistics method is obtained.Second,the up/down border and noise of the license plate are removed using texture analysis method.Third,the left/right border and noise of the license plate are removed with threshold and the accurate license plate is gotten.Image preprocessing to ensure real-time requirements is simplified.The experimental results show that the proposed method can result in the plate location accurate in real-time and can meet the system requirements.