机动车辆牌照自动定位算法研究

受光照、车型及周围景物的影响 ,将牌照从复杂背景中定位出来具有较大的难度 ,经典的定位方法不适用于此种场合 为寻求有效的解决途径 ,笔者对定位算法进行了探讨 提出了快速Bernsen二值化方法 ,用以将灰度原图二值化 采用复合极值形态滤波去除不符合牌照形状特征的线条 ,以顺序连通成分标记法标记字符区域 ,再根据牌照的纹理特征去除非法区域 ,然后对剩余区域进行模糊评判 ,选择最优 ,最终得出牌照区域 实验结果说明该算法是有效的 ,并且定位后的牌照字符完整 ,有利于下一步的字符切分和识别

A Study on Automatic Extraction of License Plates

It is difficult to extract license plates from complicated background because of illumination, automobile type and circumference. Classical waysare not applicable to this problem. To find an effective method, the author hasdived into the extraction method. A kind of binarization method called quick Bernsen is presented, it is used to make grey image binarized. By Compound Extremum Morphology Filter, lines which are not in accordance with the shape of the plate can be filtered. With the method of sequential mark, character regions are gained. Then regions which are not in accordance with the texture character of the plate are removed. Finally, the best plate region can be identified by fuzzy judge. The results of the experiment show that this method is effective; furthermore, the character regions remain integrate after extraction, so it is useful to separate and recognize characters.