复杂背景下的车辆牌照字符提取方法研究

RESEARCH ON VEHICLE LICENSE PLATE CHARACTER EXTRACTION FROM COMPLEX BACKGROUND

摘要: 针对复杂背景中字符难于提取的问题，提出了一种通用的提取字符的方法．该方法很好地利用了字符固有的特性，先对图像进行基于变分法的自适应二值化，再用一种快速区域标号算法求取连通域，然后对连通域进行处理，提取出字符行，最后进行倾斜校正．实验结果表明，该方法简单可行，成功率达到了96%以上．

Abstract: An all-purpose method for character extraction is proposed to extract difficult characters from complex background. The method uses the inherent characteristics of characters efficiently. First the image of the vehicle is binarized using adaptive thresholding. Then the connected region is obtained using fast region marking and is processed.-Finally the character row is extracted and is leveled.Experimental results indicate that the method can be implemented easily and has a success rate of over 96 percent.