摘要
提出一种基于归一化RGB和K-means聚类的车牌二值方法,实现对交通场景中的车牌阴影去除和车牌二值.首先,将RGB图像进行颜色的归一化,避免亮度改变的干扰,然后,再将图像转换到多维空间进行K-means聚类,根据聚类的标签对车牌进行二值.通过与Otsu、局部阈值等方法进行比较,该算法可以有效提高阴影覆盖车牌的二值效果.
A license plate binary method based on normalized RGB and K-means clustering is proposed to remove license plates shadows and get binary license plates in traffic scenes.By converting the RGB image to normalization space to avoid interference from brightness changing,the image is transformed in multidimensional space and classification by K-means clustering,and the binary license plate with the clustering is obtained.By comparing with the Otsu,local threshold methods,the proposed algorithm can effectively improve the binary effect of shadow covered license plates.
出处
《辽宁师范大学学报（自然科学版）》
CAS
2016年第3期349-355,共7页
Journal of Liaoning Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(41271422
61402214)
高等学校博士学科点专项科研基金资助(20132136110002)
辽宁省教育厅科学研究一般项目(L2014423)
大连市高层次人才创新支持计划项目(2015R069)
