Adaptive Binarization For Degraded Document Image Via Contrast Enhancement

In this paper, we propose an adaptive binarization method to handle different degraded document images. Firstly, a contrast image robust to background variations caused by different types of degradation is generated after preprocessing the input image. Secondly, an image contrast enhancement technique is used to strengthen the edges of the strokes to make them more complete. Then, the OTSU method is applied on the enhanced contrast image to obtain the stroke edge pixels, and the final binarization image is returned according to the threshold determined from the mean and the standard deviation of the intensity of these edge pixels. Experimental results on two public datasets indicate that the proposed method can deal with various degraded images, improve the continuity of the strokes and achieve superior binarization performance when compared with other methods.