一种新的车牌号码倾斜校正方法  

An algorithm for vehicle license plate tilt correction base on singular value decomposition

作　　者：潘梅森[1] 肖政宏[1] 

机构地区：[1]湖南文理学院计算机科学与技术系,湖南常德415000

出　　处：《光电子．激光》2008年第8期1116-1120,共5页Journal of Optoelectronics·Laser

基　　金：湖南省自然科学基金资助项目(05JJ40098);湖南省教育厅科研资助项目(05C720)

摘　　要：利用矩阵奇异值特征向量计算倾斜车牌图像的旋转角α,把倾斜图像旋转±(90-α),进行水平倾斜校正;在垂直倾斜校正中,论文提出了K-means聚类拟合直线法和最小二乘拟合直线法等两种校正方法,分别计算出垂直倾斜角θ,并对水平倾斜校正后的图像进行逆错切变换,得到最终的校正图像。实验结果表明,论文方法能快速准确地检测出图像的倾斜角,并具有较强的抗干扰性和应用适应性。相对于Hough变换校正方法,论文算法缩短了处理时间,在相同的条件下,处理时间比Hough变换快了4.13倍。为图像倾斜校正提供了一个新的有效方法。An algorithm for vehicle license plate tilt correction based on singular value decomposition was proposed. According to the feature of singular value,the rotation angle a was calculated using the matrix characteristic vector and the whole image is rotated by the angle degree ±（90-α） ,namely,image horizontal tilt correction is performed. In the vertical tilt correction processing,two correction methods of line fitting based on K-means clustering and line fitting based on least square are pro posed to compute the vertical tilt angle θ,the rotated image is done inverse shear transform and the final correction image is obtained. The experimental results show that,this algorithm can both be implemented easily,can quickly and accurately get the tilt angle, and can offer robustness when dealing with dirty vehicle license plate, in non uniform illumination conditions. This algorithm raises the computation rate and is more effective compared with Hough transform. Under the same condition, the processing time of this paper algorithm is 5.13 times faster than that of Hough transform It also provides a new effective way for image tilt correction.

关 键 词：倾斜校正 奇异值 聚类 最小二乘法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]