Shobhika T. Ingle, S. P. Bhosale
Abstract: Binarization technique for degraded document is widely used technology all over the world. Various binarization techniques are used in practices but no technique used till now is full proofs which work for all kind of degraded document. Due to the high inter intra variation between the document background and the foreground segmentation of the text from degraded document is very challenging task. In the proposed method, first of all we convert the text that is input image to black and white. And then further apply thresholdinding algorithm to the resulted document. Post processing of the resulted document is done using sobel method for edge detection and ostus algorithm for every window to achieve proper weight so that high intravarationion of the document can be detected. And then, morphological median filtering is introduced to eliminate the salt pepper noise. One more addition to the proposed method is that the OCR is added so that text is easily recognized. The experiment results show that the proposed method runs quickly, accurately and fits for all kinds of degraded document.
Keywords: Window thresholding, Classification of pixel, Sobel Binarization of document, Processing of degraded document