基于主元分析的倾斜车牌图像校正方法研究

Study on Method of Slant License Plate Correction Based on Principal Component Analysis

文章提出了一种基于主元分析(PCA)的车牌图像倾斜校正新方法。该方法将原始的像素坐标矩阵经过中心化后转换为2维协方差矩阵,再奇值分解为能反映图像倾斜方向的2维对角矩阵和坐标变换矩阵。校正算法的时间复杂度分析与实验结果均表明:相对于HOUGH变换等校正方法,PCA方法缩短了计算时间1￣2个数量级,并且在污迹、光照不均等条件下也能获得较好效果。

This paper presents a new remedy method based on principal component analysis （PCA）. The method uses license plate image data set by arranging coordinate matrix into two-dimension covariance matrix, on which centering is operated, then by applying the singular value decomposition, this matrix is refold to the bi-diagonal matrix and coordi- nate transform matrix, which are consistent with the main slant direction of the license image. The time complexity of the PCA method is analyzed in the paper. The experiment results demonstrate that the method can raise the computation rate by 1-2 orders compared with the Hough method and be effective when dealing with images of dirty vehicle license or in variant lighting conditions.