基于OpenCV的车牌图像定位

现有的车牌定位方法几乎都需要先对白天夜晚的场景进行分类,在图像中存在其他灰度剧烈变化区域时,这种场景分类容易影响到车牌定位的准确率;为了对获取到的车辆图像进行准确地定位,提出了一种改进的灰度跳变车牌定位算法,首先加入了光照补偿,使得不需要对白天和晚上的场景进行分类；然后进行了长竖线的噪声去除,可以在背景比较复杂的情况下准确定位车牌；并对精确定位出的车牌进行边框的去除,有利于下一步的字符分割;另外,对OpenCV计算机视觉库的库函数的利用,算法的复杂度得到了简化,从而更好地满足了车牌识别系统的实时性；通过对出入口处采集到的分辨率为704×576的300张图片进行测试,定位率高于95%,对白天和夜晚获取的图像有很好的适应性,满足设计要求。

A OpenCV-based vehicle license plate location

Almost all existing methods need to classify the day and night scene, and the accuracy of the positioning of the plate by this classification is easily affected when there are other dramatic changes in the gray area in the image. This paper presents an improved gray level jump License Plate Location Algorithm . First the illumination compensation is joined, makes no need for day and night scene classification. Then the long vertical lines of noise is removed, in the case of complicated background . And the frame is removed after the accurate location of license plate, which is benefited to the next character segmentation . In addition, the OpenCV library is used in the algorithm, which is made the algorithm more simplify and real-time. The accuracy is more than 95% by testing 300 images in resolution ratio of 704×576.