基于支持向量机的低质量文档图像二值化

摘要

针对低质量文档图像受墨迹浸润、页面污渍、背景纹理或光照不均等因素的影响,提出一种基于支持向量机(SVM)的低质量文档图像二值化方法.该方法对文档图像进行分块,并增强每个图像块的局部对比度;利用SVM将这些图像块分成三类,对不同图像类采用不同的阈值处理,实现粗略分割;通过笔画宽度估计确定邻域窗尺寸,从而实现局部精细二值化.实验结果表明,该算法无论从二值化图像质量,还是各种评估参数,较其他经典文档二值化方法都具有明显优势.%In view of the influence of ink infiltration,page stains,background texture or uneven illumination of low-quality document images,a low-quality document image binarization method based on Support Vector Machine(SVM)is proposed.First,the method segmented the image into blocks,and enhanced local contrast of each image block.Then, we classified those image blocks into three categories via SVM classifier.Three different thresholding were respectively used for the three categories, which achieved a rough segmentation.Finally, we determined the neighborhood window size by means of stroke width estimation, so as to achieve a precise segmentation of the foreground text and document background.Experimental results showed that the proposed algorithm had obvious advantages over other binarization methods in terms of binary image quality and various evaluation parameters.