摘要
针对光照不均的文本图像在二值化后文字识别率底下的问题,文章提出了一种针对光照不均图像进行二值化和图像增强的算法。针对图像进行分块处理,根据每块自身的平均灰度值和均方差以及全图的平均灰度值来进行算法的自适应变换,已达到对不同光照的区块进行不同处理的目的。实验结果表明本算法能够比较有效的完成对光照不均的文本图像的处理,比单纯的二值化处理方法有一定的提高。
Binaryzation can cause text images with uneven illumination to have low character recognition rates. Aiming to solve this problem, we introduce a novel algorithm designed for the binaryzation and enhancement of unevenly-illuminated text images. The method includes partitioning the images and enabling each portion to adapt transformations automatically. With the average grey levels and mean square deviation of each portion and the average grey levels of the whole picture, image areas with variant illuminations can be managed differently. Experiment results show that this algorithm can achieve the goal of such managements relatively effectively, improving the simple binaryzation method.
出处
《电脑与信息技术》
2014年第1期14-16,共3页
Computer and Information Technology
基金
十一五国家科技支撑平台重点基金项目(项目编号:2009BA171B02)
关键词
光照不均
文本图像
均方差
uneven illumination
text images
mean square deviation
