Two Thresholding for Deriving the Bi-level Document Image
Optical character recognition occupies a very important field in digital image processing. It is used extensively in daily life. If the given image does not have a bimodal intensity histogram, it will cause segmenting mistake easily for the previous algorithms of image binarization. In order to solve this problem, a new algorithm is proposed in this paper. The proposed algorithm uses the theory of moving average on the histogram of the fuzzy image, and then derives the better histogram. Since use only one thresholding value cannot solve this problem completely, the edge information and the window processing are introduced in this paper for advanced thresholding. Thus, a more refine bi-level image is derived and it will result in the improvement of optical character recognition. Experiments are carried out for some samples with shading to demonstrate the computational advantage of the proposed method.
KeywordsOriginal Image Document Image Optical Character Recognition Thresholding Method Edge Information
Unable to display preview. Download preview PDF.
- 2.Otsu, N.: A threshold selection method from gray-level histograms. IEEE Transactions on systems, Man, and Cybernetics 17, 1191–1201 (1979)Google Scholar
- 4.Niblack, W.: An Introduction to Digital Image Processing. Pretice-Hall, Englewood Cliffs (1986)Google Scholar
- 5.Liu, Y., Srihari, S.N.: Document image binarization based on texture features. IEEE Transactions on Pattern Analysis and Machine Intelligence 19, 533–540 (1997)Google Scholar
- 8.Jones, C.P.: Investments Analysis and Management. Wiley, Chichester (1991)Google Scholar
- 9.Sobel, I.E.: Camera Models and Machine Perception. PhD thesis, Stanford University (1970)Google Scholar
- 10.Bersen, J.: Dynamic thresholding of gray-level images. In: Proceeding Eighth International Conference Pattern Recognition, pp. 1251–1255 (1986)Google Scholar