基于图像的车牌缺损区域恢复识别算法研究

Study of Vehicle License Plate Recognition Method and Simulation in Adverse Weather Conditions

针对当外界天气、光照条件恶劣的时候,存在采集的车牌图像像素丢失的问题,导致无法准确识别车牌内容,文中提出一种基于改进高斯模型的残缺车牌图像像识别算法.首先对待检测图像进行亮度补偿、边缘检测、倾斜校正等预处理,通过将图像划分为若干个矩形子区域,计算各子区域的灰度平均值作为提取的图像初始特征,计算初始特征的先验概率并对后验概率进行修正,实现了对缺损区域特征值的校正,最后建立高斯模型完成车牌图像的识别,克服了传统方法无法准确识别残缺车牌图像的问题.实验证明:这种方法能够准确识别恶劣天气下的车牌图像像,取得了不错的效果.

The problem of license plate recognition in inclement weather. Traditional PCA vehicle video detection methods, when in the bad outside weather like in poor light conditions, there is loss of image pixels, leading to the problem can not accurately identify the license plate. In order to improve the accuracy of license plate recognition, this paper presents a movement license plate recognition algorithm. By the three image sequences adjacent to the vertical and horizontal directions for the poor projection, a full segmentation of images of vehicles to avoid the read only a single pixel to solve the problem of missing pixels to achieve an accurate license plate recognitioru Experimental results show that using this algorithm the adverse weather conditions on the vehicle license plate image segmentation and complete the correct identification, to obtain satisfactory results.