基于灰度方差和边缘密度的车牌定位算法

Car license plate location algorithm based on intensity variance and edge density

在车牌定位的过程中,由于光照变化、视点和距离变化、车辆运动、复杂背景等原因,摄像机很难获取到高质量的图像。为了克服这些问题对以后的车牌定位算法的影响,提出了利用车牌区域的性质来进行车牌定位的新算法。车牌区域具有在一定范围内灰度方差近似相等和边缘密度近似相等的性质,利用这些性质来增强车牌区域。为了定位车牌区域,提出基于车牌区域边缘密度性质的匹配滤波器,该滤波器可以有效地提取出所有候选目标。利用形态学和先验知识进行目标真实性验证。通过对不同环境条件下获取的700幅图像进行实验,实验结果表明该算法有效地提高车牌区域的图像质量且车牌目标定位准确率达到98.4%,验证了算法的有效性和鲁棒性。

In the process of car license plate location,camera can＇t obtain high quality images easily due to illumination conditions,viewpoint and distance changes,vehicle motion,complex background,and etc.In order to alleviate the impact of these problems on car license plate location algorithm,a new algorithm using the properties of license plate region is proposed.The license plate region has the properties that the intensity variance is approximately equal and edge density is also approximately equal and within a certain range,which are used to enhance the license plate region.In order to locate the license plate region,a matched filter based on the property of edge density is designed.The filter can effectively extract all the possible objects.Then,the object authenticity confirmation is performed using morphological operation and priori knowledge.An experiment was carried out with 700 actual images acquired under different environments,and experimental results show that the proposed algorithm can effectively improve the image quality of license plate region and the object location accuracy is 98.4%,which proves that the proposed object location method is effective and robust.