复杂背景下的混合特征车牌定位方法

针对各种复杂背景的车牌定位问题,提出一种复杂背景下基于车牌混合特征的车牌定位算法。首先对彩色图像进行预处理,并利用基于边缘检测方法进行二值化;然后结合横向数学形态学运算和车牌几何形状特征,提取出矩形车牌候选区域;最后根据车牌颜色特征在HIS空间下结合垂直和水平投影对车牌区域进行精确定位。实验表明,该算法适用于任意大小、位置和背景环境下的车牌定位,能有效解决仅仅依靠纹理信息或颜色信息车牌定位率低的问题,具有较强的鲁棒性。

Approach for License Plate Location Based on Mixed Features under Complex Background

Proposes a novel locating approach to extract the number plate from the complex background which based on the mixed features of vehicle license.Firstly,color image should be preprocessed and binarization is done based on edge detection.Then gets the rectangular localization candidate regions by the horizontal mathematical morphology operation and geometry characteristics of the license plate.According to the color feature in the HIS space,uses horizontal projection and vertical projection to locate the real license plate area from candidate regions.Experiment shows that this method can be applied to any car size and position and image background,and can effectively solve the inefficient of license plate location only by texture or color information,has strong robustness.