摘要
对较厚的书籍进行扫描时,由于中缝难以压平,常使扫描后的图像出现黑边,因而采用常规的算法对其进行阈值化结果很不理想,进而严重影响到文字识别的效果。针对这一问题,提出了一种基于纹理分析的图像二值化方法。该方法在动态阈值的基础上,通过对分块后子图像的纹理特性进行分析,将子图像分成不同的类型,进而对不同类型的子图像进行不同的处理,很好地消除了图像黑边和背景亮度变化对二值化结果的影响。
Black pieces always appear on the image margin while scan the thicker book,because the middle gap of thicker book is hard to planish.The ordinary threshold method can＇t achieve the perfect effect,and it will affect character recognition.This paper proposed a new binarization method to scanning image.Based on dynamic threshold,used texture analysis on partitional images,then separated partitional images into different types,and different types would have different algorithms to cope with.This method avoids influences brought by black pieces and the change of background luminance.
出处
《计算机应用研究》
CSCD
北大核心
2011年第5期1998-2000,共3页
Application Research of Computers
关键词
二值化
书籍扫描图像
动态阈值
图像分块
纹理分析
binarization
book scanning image
dynamic threshold
image partition
texture analysis
