An Improved Algorithm for Document Image Enhancing

Document image from a digital camera or a digital scanner always shows different types of distortions. Therefore, we must perform some corrections to the image object before reading the text information with an OCR engine. To improve the distorted document image's vision quality and the OCR rate, this paper proposed an improved algorithm for document image enhancing by recovering the deformed rectangle object. Firstly, we detect the object to be corrected for skew and perspective distortion. Secondly, we measure the size of the detected object to project the data on the exact real object-size. Third, we divide the entire image to find the position of some specific object points. Finally we deform the pixel grid and map this deformed grid to the destination image to produce the final corrected image. As a result, for the corrected image, the distortions are mostly removed leading the OCR to get much better results. Experimental results show that our approach can efficiently resolve the issue of document image distortion mainly for perspective distortion.