基于色彩纹理的车牌定位系统设计

摘要

To improve the accuracy of license plate location,we proposed a new method of license plate location based on color texture. First,we converted the images from red,green and blue space to hue, saturation and value space to generate three component images. Then we selected the gray values of effective pixels matching the license plate region to eliminate the unuseful information in image processing after the images were filtered. Two binary images were generated by using the color of license plate background and characters for image binarization. Scanning and comparing the corresponding pixels progressively in the two binary images,we can judge whether the region matching the license plate texture is found by counting gray level jump times,then determine and display the region of the license plate in the image by calculating the amount of effective pixels. This method can significantly improve the accuracy and stability of the recognition of the vehicle license plate.%为了提高车牌定位的准确率,提出了一种基于色彩纹理的车牌定位的分析方法.首先将彩色图像的色彩空间由RGB转换到HSV,生成HSV色彩模型的三通道图像,将图片进行滤波调整之后,并将符合车牌区域的有效像素的灰度值范围作为参数排除图像中的干扰信息,然后将转换后的图像车牌背景颜色和车牌字符颜色进行二值化处理生成两幅灰度图像,采用逐行扫描的方法对两幅灰度图像的各个像素点进行分析和比对,通过像素灰度值的跳变次数,判断是否找出符合车牌纹理的区域,通过计算确定车牌在图像上的区域,并输出车牌图像.该方法提高了的车牌识别的准确性和稳定性.