摘要
准确分割车牌图像,可为分割目标的提取奠定良好的基础。试验中采用基于OTSU和区域生长的算法,实现了车牌图像的精准分割,通过最大类间方差法得到自适应阈值,代替了传统手动选择阈值的方法,再结合中值滤波对图像进行预处理,避免了传统阈值造成的过分割或者欠分割现象,且预防图像噪声,使得图像像素值变得缓和,从而有效地分割出车牌图像。
New license plate segmentation technologies are emerging. In order to accurately segment the license plate image, which can lay a good foundation for the extraction of the target, the paper presents a test of the algorithm based on OTSU and regional growth. The test realized the precise segmentation of license plate image, obtained the method of adaptive threshold replacing the traditional manual threshold selection method by Otsu. It then preprocessed the image combined with median filter to avoid the traditional threshold caused by the over-segmentation and under-segmentation phenomenon.Results showed that it is effective in the prevention of image noise, and the image pixel values become more relaxed, so as to effectively segment the license plate image.
作者
吴聪
殷浩
黄中勇
王凯
WU Cong YIN Hao HUANG Zhongyong WANG Kai(School of Computer Science, Hubei Univ. of Tech. 430068 Wuhan, China China Construction Third Engin.Bureau. Installation Engin. Co. Ltd, Wuhan 430064, China)
出处
《湖北工业大学学报》
2017年第2期58-61,共4页
Journal of Hubei University of Technology
基金
国家自然科学基金青年项目(61300127)
湖北省自然科学基金项目(2012FFB00701)
