基于数学形态学的车牌定位算法

License Plate Location Algorithm Based on Mathematical Morphology

在汽车牌照识别系统中,车牌定位是整个识别模块实现的前提,为了能对汽车牌照精确定位,提出了一种基于数学形态学的车牌定位方法.首先利用中值滤波方法对汽车图像进行消噪处理,然后用迭代阈值选择法将图像进行二值化,并运用数学形态学的膨胀、腐蚀对二值图像进行处理,得到几个车牌候选区,然后利用面积、长宽比以及垂直投影特征值等进行综合分析,准确定位车牌区域.实验结果表明,该方法简单易行、准确率高、并且具有一定实效性,可用于对实际车牌图像的准确定位.

License plate location is the premise of the whole recognition module in the system of license platerecognitions. In order to be able to pinpoint the car license plate, this paper proposes a license plate locationmethod based on mathematical morphology. Firstly, utilizing the median filter to eliminate noise of the carimage, then using iterative threshold selection method to binarize the image, and the binary image isprocessed by means of mathematical morphology expansion and corrosion to get the license plate numbers ofthe candidate area. The comprehensive analysis of license plate area and aspect ratio and vertical projectioneigenvaluesis applied to accurate positioning plate area. The experiment turns out that the method is simpleand easy with high accuracy and effectiveness, which can be used to accurately locate the actual license plateimage.