基于NSCT变换的车牌定位算法研究

Car License Plate Location Algorithm Research Based on NSCT Transformation

在汽车牌照识别系统中,车牌定位是整个识别模块实现的前提,目前车牌定位的方法多种多样,各有所长,但存在着计算量大或定位准确率不高等问题。为了实现对车牌区域的精确定位,提出了一种基于非下采样Contourlet变换的车牌定位算法。首先,对图像进行非下采样Contourlet变换,得到车辆图像的8个方向的高频分量子图;然后,通过一定的结合规则将这些高频子图合成一幅能突出车牌区域的高频图;最后,运用数学形态学和连通域分析定位出车牌。实验结果表明,其算法不仅能成功提取车牌图像边缘,而且能很好地滤除噪声,从而实现准确车牌定位。

In car license plate identification systems, license plate positioning is the precondition of the whole identification module. Now various methods are applied and each of which has its own advantage. However, problems such as large amount of calculation or the low positioning accuracy are still remained. To implement accurate positioning of license plate, an algorithm of license plate location based on Nonsubsampled Contourlet Transform （NSCT） is presented. Firstly, the license plate picture is processed with Nonsubsampled Contourlet Transform （NSCT） to get 8 oriented high-frequency components pictures of the original picture. Next, the high frequency subgraphs are merged into one High frequency chart which can highlight the plate through a certain combination role. Finally, the license plate is positioned with mathematical morphology and connected components analysis. The experimental results indicate that the algorithm can not only extract the image edge of car license plate, but also can greatly reduce the noise. It is able to implement car license plates＇ accurate positioning.