Document image preprocessing based on optimal Boolean filters

In this paper, optimal Boolean filters are applied to enhance the binary do cument images corrupted with uniform noise or uniformly distributed distinc t graphical patterns in the background. The performance and operation theor y of optimal Boolean filters against other competitive techniques are compa red, Experimental results show that the Boolean filters outperforms the mor phology approach in extracting the text from overlapped text/background ima ges, The feasibility of trained Boolean filters is also confirmed by experi mental results in the case where the original image is not available. (C) 2 000 Elsevier Science B.V. All rights reserved.