基于背景规范化的照相文本图像二值化

Binarization of Photo Document Image Based on Background Normalization

提出一种基于背景规范化的照相文本图像二值方法.通过方差计算确定文本图像的文字区域,并对文字区域进行背景插值后得到原始文本图像的背景图.根据背景图把原始文本图像转换成背景规范化的文本图像,同时用拟合曲线对不同明暗背景区实行相应的对比度补偿,把前景像点与背景像点设置到较小的灰度范围,然后用全局阈值方法进行二值化处理.试验结果表明,该方法是有效的.

A method of photo document image binarization based on background normalization was pro- posed. By calculating the variance,the text area of an document image was located and the background image of the original document image was obtained by background interpolation to the area. Then the original document image was transformed into a background normalized image and contrast compensation of light and dark background areas was implemented by fitting curve. By setting pixels of foreground and background to small gray-scale, the binarization was achieved with global threshold method. The experimental result shows that the new method is efficient.