基于背景和像素的彩色车牌定位

Color license plate location based on background and pixel

提出结合车牌的背景差和像素分类实现彩色车牌定位算法。首先采用背景差法,即原图与背景图像相减,去除大量疑似车牌颜色特征的背景区域,缩小定位范围;然后统计车牌颜色RGB值,利用像素分类对车牌进行水平和垂直精确定位,确定车牌区域。实验表明,该方法定位准确率高,适用于任意背景、位置和光照下的车牌定位,明显克服了用车牌单一特征分析时难以解决的车牌区域变大的问题。

In this paper, the background difference with license plates and pixel classification to achieve color license plate location algorithm. First, using the background difference method, namely, Original and background image subtraction to remove the license plate color characteristics of a large number of suspected background area, narrowing down; and then statistics plate color RGB values, using pixel classification of license plates in horizontal and vertical precision position to determine the license plate area. Experiments show that positioning with high accuracy, the method is suitable for any background, location and illumination under the license plate location, obviously overcome the characteristics of the analysis with a single license plate license plate regions are difficult to resolve larger issues.