面向自然环境的车牌检测算法设计

Design of License Plate Detection Algorithm for Natural Environment

作为现代化智能领域中的主要技术,车牌的自动识别理论中主要包括了两个不同的部分,分别是车牌的检测与字符的识别.根据车牌检测算法的发展现状,文中出于对车牌检测速度与精度的综合考虑,采用了一种基于学习、由粗检测到精检测的车牌检测技术与方法.该方法在利用车牌的颜色信息与边缘信息对车牌候选区域进行快速检测的基础上,通过数字车牌图像中的梯度方向直方图与支持向量机来实现对真实场景中车牌候选区域的高效筛选.该方法能够对多幅不同光照条件、不同拍摄角度、不同拍摄时间所获取图像进行检测,实验结果表明该算法具有精度高、鲁棒性好的特点,可以满足自然条件下对车牌进行检测的应用需求.

As the main technology in the field of modern intelligent efficiency,the theory of auto recognition of license plate mainly includes two different parts,which are the detection of license plate and the recognition of characters.According to the development status of the license plate detection algorithm and out of the comprehensive consideration of the speed and precision of the license plate detection,a kind of license plate detection technology and method based on learning,from rough detection to fine detection are adopted in this paper.On the basis of using the color information and the edge information of the license plate for the fast detection of the candidate region of the license plate,the gradient direction histogram and SVM in the digital license plate image are used to efficiently screen the candidate regions in real scene.The method can detect the images obtained with different illumination conditions,different shooting angles and different shooting time.The experimental results show that the algorithm has high accuracy and good robustness and can meet the application requirements for the detection of the license plate in natural conditions.