基于颜色对特征点主成分分析的车牌校正方法

Slant Correction of Vehicle License Plate Based on Color Pair Principal Component Analysis

为了在车牌的倾斜校正过程中减少车牌边框和噪声干扰的影响,并简化计算过程,提出了一种基于颜色对特征点主成分分析(PCA)的车牌水平校正方法。该方法根据车牌背景与字符交界处的颜色具有固定搭配这一特点,首先在原始车牌图像中提取颜色对特征点,并将所有颜色对特征点视为待分析的样本;然后构建这些样本特征点的2维散布矩阵,并通过主成分分析求出其主成分方向,该主成分方向就是车牌的水平倾斜方向;最后再进行相应的旋转,即可获得校正后图像。由于这种方法将车牌的颜色信息和边缘信息融合后共同使用于车牌的校正过程,同时将搜索图像倾斜角度转换为2维实对称矩阵进行计算,从而简化了计算。通过编程对实际车牌图像进行的实验结果证明,该方法对于边框不清或含有噪声干扰的图像仍然能取得较好的校正结果。

In this paper ,slant correction methods of vehicle license plate are analyzed and an approach based on color pair and principal component analysis is prestented. A pixel is considered as a color pair point if its color pattern matches the combination of the background color and text color of the plate. The principal component orientation of the plate is achieved through principal component analysis of the color pair pixels. The principal component orientation is considered as the slant angle of the plate and the correction of the plate is accomplished. The experiment results demonstrate that this method makes the correction of the plate easier and more precise. The images of dirty vehicle license can also obtain effective results through this algorithm.