车牌识别中的预处理包括图像增强,二值化和噪声去除。由于车牌图像摄取的特殊性,使得车牌在图像中通常是倾斜的,而倾斜的车牌是很难被分割和识别的,因此车牌倾斜矫正已成为车牌识别系统中一个必需解决的问题。给出了基于图像特征线的二值化阈值确定和针对车牌图像的倾斜矫正改进算法,对车牌的水平和垂直方向分别采用了不同的矫正方法,实验结果表明此算法是快速有效的。
The preprocessing algorithm for License Plate Recognition (LPR) includes image enhancement, binary conversion and noise reduction. Due to the particularity of shooting license plate,it is usually slantwise in image, causing the difficulty of segment and recognition,therefore,slant correction of license plate must be carried out in LPR system. In this paper a binary conversion method based on feature line is proposed and an improved algorithm of slant correction of license plate is given,where different algorithms are used for horizontal and vertical directions. Experimental results are provided to demonstrate the performance of the proposed algorithm.