基于边缘与颜色信息的车牌精确定位算法

针对现有车牌定位算法定位率不高和速度慢等问题,结合数学形态学,提出了一种基于车牌边缘检测与颜色信息的精确定位算法.首先对图像进行灰度化、平滑去噪及灰度拉伸等预处理,然后对图像进行垂直边缘检测和二值化处理;再采用线过滤算法滤去干扰边缘线信息,并利用车牌区域灰度变化次数多的特性定位出车牌上下边缘;对经过上下边缘定位后的车牌采用点密度过滤算法滤去零散点,接着采用数学形态学方法寻找连通域粗定位出车牌区域;最后对粗定位车牌在HSV空间进行颜色分割和倾斜矫正,从而精确定位出车牌.实验结果表明,该算法能够实现车牌的快速精确定位.

Precisely license plate locating algorithm based on edge and color information

A precise license plate locating algorithm was proposed combined with the mathematical morphology to solve the problems of existing locating algorithms such as low locating rate and slow speed. The algorithm was based on edge detection and color information. First color-image was preprocessed with grayscaling, smoothing and gray stretching and so on, then vertical edges detection and binarization were implemented on the image. After that linear filtering algorithm was adopted to filter disturbing edge lines, and up-down plate edges were located utilizing the characteristic of plate area′s variety in gray scales. Isolated points of the up-down edges located plate were filtered by using a point density filtering algorithm, then the plate area was located coarsely by finding connected components with adopting the mathematical morphology. Finally the plate was precisely located through color segmentation in HSV space and incline rectification. The experimental results show that this algorithm can locate the plate precisely and quickly.