Recovery Of Degraded Image Documentation By Using Image Binarization Method

Now a days, many activities which depend upon internet. And there is a great need to move all these activities which are implemented by end user towards digitalization of world. Every times it happens that institute, firm and organization have to maintain the books or novels for longer time span and there occurs a new challenge for institutes. Books are the physical entity so it will definitely have the issue of wear ad tear. The text on pages gets degraded because pages are affected by external environment. The data on the pages can be confidential and sensitive and there should be very robust and dynamic mechanism for preserving the data on same. Due to this degradation many of document images not in readable form. So, there is a need to separate out text from those degraded images and preserve them for feature reference. This gives great reason for developing a foreground text extraction mechanism that will aid in preserving the document or in other words, the text on those documents. The proposed system includes such a mechanism that not only helps to detect textual matter on the document but also preserve the text on the other image. Previously, many such algorithm have been proposed for this purpose, but as seen by research done for years, optical character recognition (OCR), Handwritten text recognition such algorithm were developed but there are still few areas which were yet to be worked on. The proposed system focuses on improving the text extraction efficiency and therefore eradicates the use of Canny’s edge map and makes use of simple Otsu thresholding and edge detection and luminance grayscale method for improving the detected edge sharpness. Also the very important aspect on text extraction is clarity of text being extracted. In this paper, post processing algorithm works on the same task for smoothing the extracted text and also removing the unwanted pixels from the image. This algorithm includes image contrast inversion, edge estimation, image binarization and post processing of binary image. We can able to separate out the foreground text from background degradation after applying these all methods.