摘要
研究图像二值化分割问题。针对模糊或者蜕化文档图像背景与文字融合在一起导致难以区分的难点,提出一种快速有效的两级结构图像分割算法。首先利用迭代算法对图像进行单一阈值分割,在每次迭代过程中以图像均值为依据,对图像进行均衡化处理;在基于全局分割的基础上,在局部范围内根据噪声的统计特性对文档图像进行去噪处理。方法简单高效,实验结果显示该方法能快速地将文档中的文字与背景进行分离,为后续的文档自动化处理提供准确有效的二值化图像。利用该方法,可以方便地拓展到其他类型的二值化处理系统中,例如车牌分割等。
In this paper,we discuss image＇s binary segmentation issue.In light of the hard case that the integration of the blurred or degenerated document image background with the texts causes difficulty in differentiation,we put forward a fast and effective segmentation algorithm for images with two-level structure.First,the iteration algorithm is employed to segment the image for single threshold value,in each iteration process the image mean is used as the base to carry out the equalisation processing on the image;on the basis of global-based segmentation,denoising processing is conducted on document images within local scope according to statistical features of the noise.The method is simple and effective.Experimental result demonstrates that the algorithm in the paper can quickly separate the texts from the background in a document,this provides a precise and valid binarised image to following automated document processing.This method can be conveniently promoted to binarisation processing system of other types such as vehicles licence segmentation and so on.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第9期246-248,292,共4页
Computer Applications and Software
关键词
文档图像
二值化
图像处理
图像分割
Document image ,Binarisation ,Image processing, Image segmentation
