复杂背景下车牌的定位和字符分割技术研究

Technology Research of Vehicle License Plate Location and Character Segmentation Under Complex Scenes

提出了一种基于数学形态学的多特征车牌定位方法和基于模糊模板匹配及垂直投影的字符分割算法,先通过形态学运算得到一系列候选区域,根据车牌的纹理特征从中找出车牌区域,再利用模糊模板匹配的方法找到字符区域,进而根据字符垂直投影进行单个字符分割。对大量的图片进行实验,结果表明该算法能够有效地解决复杂背景下车牌定位和字符分割困难的问题,具有较好的鲁棒性。

An approach of vehicle license plate location based on morphology and Multi-characteristics of license plates and a vertical projection-dim template matching algorithm for character segmentation are proposed in this paper, some candidate regions can be obtained by morphological operations, then genuine region will be determined by the textures of car license plates, next find the character region by dim template matching, finally use vertical projection to deride the single character. The results demonstrate the effectiveness of proposed algorithms under complex scenes by the experiments with a large set of car images, and the algorithms have high robustness.