基于梯度与形态学的低质量车牌定位

针对低质量车牌的定位问题，提出一种水平梯度信息和数学形态学操作相结合的定位方法。利用自适应梯度图阈值分割算法分割出车牌部分的梯度信息；通过形态学和区域合并操作得到连通的候选区域；结合车牌自身特点进行区域搜索，粗定位出车牌；用扫描线法去除边框，实现车牌的精确定位。实验结果表明，该方法对光线昏暗、噪声严重、车牌区域污损、倾斜等条件下的车牌定位具有较强的鲁棒性，能满足实际交通场景中的车牌定位需求。

Low-quality License Plate Location Based on Gradient and Morphology

A new method based on horizontal gradient and morphology is proposed for the location of low-quality license plate.An adaptive threshold segmentation algorithm for horizontal gradient image is used to obtain useful gradient information of the image,then the adjacent candidate regions are obtained by morphological operation and region merging.The rough location result of the license plate is obtained by search candidate regions according to the characteristics of the plate.The accurate region of the plate is obtained by the line scanning method.Experimental result shows that the algorithm is robust for the location of low-quality plates taken under the environment of low illumination,serious noise,slanting plate or partial pollution and can meet the demand of real-time traffic