笔划宽度变换直方图的车牌识别中的应用

对于很多车牌识别中使用的算法而言,主要存在两种车牌类型：深色文字浅色背景车牌和浅色文字深色背景车牌。这两种车牌主要是二值化结果不同,进而导致后续字符分割和识别处理的图像类型不同。因此,判断车牌的二值化类型对于车牌识别来说是基础且重要的工作。本文提出了一种基于字符笔划宽度变换直方图的二值化算法,根据正色图像和反色图像计算比较两者笔划宽度变换直方图的最大值来判别车牌类型。同时,根据判别结果还可以估计笔划宽度,为局部二值化算法的邻域窗口大小的选择提供依据。我们使用多样式的美国车牌作为实验对象,与其他算法相比,实验结果表明该算法具有更好准确率,但也有更高的复杂度。

Application of the histograms of stroke width transform in the license plate recognition

For many algorithms used in the license plate recognition field, there are two categories of license plate: dark characters on the light background and light characters on the dark background. The main difference of them are the binarization results, and it will lead to different results of character segmentation and recognition. Therefore, the binarization type judgment is essential and important work for the license plate recognition. We propose a method based on the histograms of the stroke width transform for the problem in this paper. The proposed method judges the license plate type via comparing the maximization of the histograms of the stroke width transform of normal and inverted images. Besides, we also can estimate the stroke width by the judgment results, and it can be used for selecting the neighborhood window size of local binarization algorithms. The experimental results show that the proposed method is superior to others by using multi-style license plates in the United States, but the time complexity is high.