Abstract: According that the number of fractal dimension is related to the complication of the image,the gray’s value scope of image can be gotten by calculating the fractal dimension of the break of the object and then the threshold can be constituted.The experiment result proves that threshold value through fractal dimension can achieve excellent effect compared with traditional method.Furthermore,the character of license plate can be recognized in different sunlight by this method,and slantwise character can be distinguished from the complicated background as well.

摘要： 针对当前常用车牌识别二值化算法存在的问题,提出了基于分形维数的二值化的方法。根据分形维数反映图像复杂程度的定义,通过计算两次突变的分维数,来确定图像的灰度值范围,并利用该灰度值范围确定阈值。并通过实验,表明利用分形维数所得到的阈值进行二值化处理较传统方法有较大改进,且该方法解决了在自然光和不同光照背景下对车牌识别的干扰问题,也可以从复杂背景中提取出倾斜的车牌。