A number of techniques have previously been proposed for effective thresholding of document images. In this paper two new thresholding techniques are proposed and compared against some existing algorithms. The algorithms were evaluated on four types of 'difficult' document images where considerable background noise or variation in contrast and illumination exists. The quality of the thresholding was assessed using the Precision and Recall analysis of the resultant words in the foreground. The conclusion is that no single algorithm works well for all types of image but some work better than others for particular types of images suggesting that improved performance can be obtained by automatic selection or combination of appropriate algorithm(s) for the type of document image under investigation.
展开▼
