一种基于形态学结构元素优化的车牌快速检测算法

A Fast License Plate Detection Algorithm Based on Morphological Structural Elements Optimization

车牌检测(License Plate Detection,LPD)是自动车牌识别系统中(Automatic License Plate Recognition System,ALPRS)的重要一环.为了提高车牌检测准确率和降低检测时间,提出一种基于形态学结构元素优化的车牌快速检测算法.该算法首先对检测图像进行预处理,然后运用基于灰度图像内部结构元素(Structural Element,SE)比较的形态学Top-hat变换来突出车牌区域以及最大类间方差阈值化来降低背景干扰,并且利用形态学开操作和闭操作对不连续的椒盐噪声进行过滤,最后使用轮廓检测和几何条件检测出车牌区域.为了适应多种类型的车牌形态,该文对3种形态学操作的SE参数进行优化.实验结果表明,本文方法的检测准确率、召回率和检测时间均优于其他车牌检测方法.

License plate detection(LPD)is an important part of automatic license plate recognition system(ALPRS).In order to improve the accuracy of license plate detection and reduce the detection time,a fast license plate detection algorithm based on the morphological structural elements optimization has been proposed.The algorithm first preprocesses the detected image,and then uses the morphological Top-hat transform based on the comparison of the internal structural elements(SE)of the gray image to highlight the license plate area,reduces background interference by threshold of maximum inter-class variance,filters discontinuous salt and pepper noise by morphological opening and closing operations.Finally detecting the license plate area by contour detection and geometric conditions.In order to adapt to various types of license plate shape,the SE parameters of three morphological operations have been optimized in this paper.The experimental results show that the detection accuracy and detection time of this method are better than other license plate detection methods.