Abstract: This paper presents an effective algorithm for vehicle license plate location，which employs multi-color models.Based on the RGB and HSV color models，the algorithm makes full use of the color information to segment the color image and construct a multi-scale gray image，and then some regions with the color of car-plate are obtained.To the segmented image，the statistical-letter frequency properties and geometric structure properties are analyzed to locate the license plate correctly.Experiments show that this method is accurate and robust in locating vehicle license plates，and can be applied to any car size and position，and the image background.

摘要： 提出了一种基于多颜色模型的车牌定位新方法。该方法在RGB和HSV两种颜色模型中，首先充分利用车牌图像的彩色信息对车牌图像进行颜色分割，将彩色图像转化为多级灰度图，从而确定出车牌所在的可能区域，完成车牌区域的粗定位。然后结合车牌特有的字频统计特征和几何结构特征进行分析和判断，进而精确定位出车牌区域。实验表明该方法定位准确率高，且适用于任意背景、位置和光照下的牌照定位，具有很强的鲁棒性。