基于连通区域的复杂车牌的字符分割算法

摘要

为克服车牌识别过程中，字符分割算法受到伪车牌、阴影和倾斜等因素的影响，提出基于连通区域标记的复杂车牌的字符分割算法。经预处理消除车体和阴影对图像二值化的干扰；对被标记区域的高度和间隔特征进行特征筛选，去除复杂边框和伪车牌等因素对分割的影响；采用模板和投影法补充分割，解决字符缺省、粘连等问题。大量实例验证结果表明，该算法能有效分割复杂车牌，去除伪车牌，有较高鲁棒性和识别率。%To reduce the impacts of pseudo-plates,shadows and skewness on character segmentation in license plate recognition process,a method of character segmentation based on connected component labeling for complex license plates was presented. The interference of vehicle colors and shadows in image binarization was eliminated through image pretreatment.To avoid the effects of complex frame and pseudo-plates on segmentation,the labeled components were selected by their characteristics of height and interval.Based on the license plate template,the character segmentation was improved by projection to solve the problems of missing and connecting characters.The ability of segmenting the complex license plates and eliminating pseudo-plates with strong robustness and high accuracy is verified by experiments.