机构地区: 桂林电子科技大学
出 处: 《电视技术》 2012年第1期113-117,共5页
摘 要: 针对复杂环境下车牌定位问题,提出基于微分算子的方法。首先用二阶微分算子对图像在3×3邻域内进行处理,并基于欧氏距离变换,改进处理后的投影曲线;然后,通过曲线的波谷微分值,初选车牌区域,并根据区域内的波谷微分值比、曲线占空比等参数,对车牌在竖直方向定位;最后,在此基础上,通过水平方向二阶微分投影曲线,提取车牌。通过研究表明,该算法具有较高的准确率。 The algorithm of differential operator is proposed, which aims at license plate localization under complex backgrounds. Firstly, projection curve is improved based on Euclidean distance transform,and the image is processed by second order differential operator in 3 × 3 regions. Secondly,vertical position of license plate is located based on ration of differential value in valleys and duty factor of curve in candidate regions ,after choosing differen- tial value of valleys. At last,license plate is located by second derivative curve of the horizontal projection on the basis of vertical localization. The re- search shows that the algorithm has higher accuracy.