期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2014
卷号：5
期号：2
页码：1759-1763
出版社：TechScience Publications
摘要：Segmentation of text from poorly documented images is a very difficult task due to high mutation between the document background and foreground text of various document images. In this paper, a binarization technique is significantly designed for historical document images. This existing binarization technique points either on finding an appropriate global threshold for each area in order to remove strains, smear and uneven illuminations. In binarization process an adaptive contrast map is first constructed for an input degraded document image. Adaptive image contrast is a combination of local image contrast and local image gradient. This contrast map is then binarized and combined with Canny’s edge map to detect text stroke edge pixels. The document is further segmented by local threshold that is estimated based on the intensities of detected text stroke edge pixel within that local window. This method is simple, robust and includes minimum parameter tuning. Our approach applies a global threshold and detects image areas that are more likely to still contain noise. Each of these areas is reprocessed separately to achieve better quality of binarization.
关键词：global threshold; local image contrast; local image;gradient; Canny’s edge map; adaptive image contrast; text;stroke edge pixel
