Document Image Binarization Based on Stroke Enhancement

This paper proposes a novel document image binarization approach based on stroke neighborhood enhancement. First, foreground pixels are initially labeled. Then, the strokes are enhanced based on their neighborhood information which include gradient information and foreground-background pixel distances of foreground pixels and their neighboring background pixels. At last, the enhanced image is finally binarized. This approach can provide good result for those document images suffering from lighting variance, low resolution and blurring