Abstract: As existing vehicle license plate location algorithms tend to have a lower correctness rate on images with high resolution and complex background, this paper proposes a novel method to vehicle license plate location based on the edge detection and color-texture histogram. The proposed location method is divided into two phases. Firstly, the method extracts the vertical edges in the image by using an improved edge extraction method which combines Canny and Sobel algorithms. Then a filter and projection method is used to locate the potential plate area. To locate the plate precisely, the color-texture histogram feature of the potential region is extracted and matched to the trained classifiers. The experimental results show that, this method has a good robustness against complex backgrounds and uneven light, and can achieve good performance during both day and night.

Key words: vehicle license plate location, edge detection, color-texture histogram, support vector machine (SVM)

摘要： 针对已有车牌定位算法在分辨率高、背景复杂图像上存在准确率下降的问题，提出了一种基于边缘检测和颜色纹理直方图的车牌定位算法。该定位算法分为两个阶段：首先利用结合了Canny和Sobel算法的改进边缘检测算法提取图像的垂直边缘，并结合滤波、投影等方法进行车牌粗定位；然后提取候选区域的颜色纹理直方图，与训练好的分类器进行匹配，实现车牌的精确定位。实验表明，该方法对于背景复杂、光照不均等情况均有良好的鲁棒性，在白天和晚上都能取得较好的定位效果。

关键词: 车牌定位, 边缘检测, 颜色纹理直方图, 支持向量机（SVM）