一个改进的车牌号码倾斜校正方法

摘要

Vehicle license plate tilt correction is a very important link in the intelligent transportation system. In this paper, based on the weighted least square method, the vehicle license plate region is first fitted to a straight line, and then the line slope is calculated, from which the rotation angleαis derived. Finally the whole image is rotated by−αand image tilt correction is carried out. The experimental results reveal that, this proposed method can quickly and accurately obtain the tilt angle and has better robustness and adaptability. Compared with fitting using the least square method, the tilt angle using the proposed method is more precise and the value of objection function is less. Also, compared with Hough Transformation, this proposed method is characterized by faster processing time and more precise tilt angle, which particularly caters to the intelligent transportation system%　　在智能交通系统中，车牌号码倾斜校正是预处理的重要步骤。论文使用加权最小二乘法，把车牌区域拟合成一条直线，然后求得该直线斜率，并计算倾斜车牌图像的倾斜角α，最后把图像旋转−α，完成倾斜校正。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，并具有较强的适应性。与最小二乘法拟合直线比较，倾斜角精度更高，目标函数值更少；与Hough变换校正相比较，论文算法处理时间更短，倾斜角测量精度更高。