An efficient preprocessing block for the middle-age Persian manuscripts

In this paper, a preprocessing block for the middle-age Persian documents is proposed. The main idea is based on the mathematical morphology, connected components and clustering. The proposed algorithm is capable to simultaneously remove the noise and segment the manuscript to its basic components i.e. lines, words and characters. The proposed strategy has been tested on 200 page of the middle-age Persian. We have also used the success of the k-means algorithm on page to line segmentation as a criterion for the performance evaluation on the test data. The results show the proposed algorithm has 98.12% accuracy on page to line segmentation