Efficient adaptive thresholding algorithm for in-homogeneous document background removal
Image binarization refers to convert gray-level images into binary ones, and many binarization algorithms have been developed. The related algorithms can be classified as either high quality computation or high speed performance. This paper presents an algorithm that ensures both benefits at the same time. The proposed algorithm intelligently segments input images into several different sized sub-images by using a Sobel like matrix. After which each sub-image will be classified into background set or foreground set according to it’s feature. Finally the foreground set sub-images will be binarized by Otsu’s method independently. Experimental results reveal that our algorithm provides the appropriate quality with the medium speed.
KeywordsDocument image analysis Document image binarization Adaptive thresholding High speed Low computational cost
- 1.Beucher S (1994) Watershed, hierarchical segmentation and waterfall algorithm.. In: Mathematical morphology and its applications to image processing, Springer, pp 69–76Google Scholar
- 2.Cheriet M, Said J, Suen C (1995) A formal model for document processing of business forms. In: Proceedings of the Third International Conference on Document Analysis and Recognition, vol 1, pp 210–213Google Scholar
- 5.Hegt H, Haye R, Khan N (1998) A high performance license plate recognition system. IEEE Int Conf Syst Man Cybern 5:4357–4362Google Scholar
- 7.Otsu N (1979) A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern SMC-9:62–66Google Scholar