Document Image Dewarping Using Deep Learning

Vijaya Kumar Bajjer Ramanna, Syed Saqib Bukhari, Andreas Dengel

In: The 8th International Conference on Pattern Recognition Applications and Methods. International Conference on Pattern Recognition Applications and Methods (ICPRAM-2019) February 19-21 Prague Czech Republic Insticc 2019.


The distorted images have been a major problem for Optical Character Recognition (OCR). In order to performOCR on distorted images, dewarping has become a principal preprocessing step. This paper presents a newdocument dewarping method that removes curl and geometric distortion of modern and historical documents.Finally, the proposed method is evaluated and compared to the existing Computer Vision based method. Mostof the traditional dewarping algorithms are created based on the text line feature extraction and segmentation.However, textual content extraction and segmentation can be sophisticated. Hence, the new technique is pro-posed, which doesn’t need any complicated methods to process the text lines. The proposed method is basedon Deep Learning and it can be applied on all type of text documents and also documents with images andgraphics. Moreover, there is no preprocessing required to apply this method on warped images. In the pro-posed system, the document distortion problem is treated as an image-to-image translation. The new methodis implemented using a very powerful pix2pixhd network by utilizing Conditional Generative AdversarialNetworks (CGAN). The network is trained on UW3 dataset by supplying distorted document as an input andcleaned image as the target. The generated images from the proposed method are cleanly dewarped and theyare of high-resolution. Furthermore, these images can be used to perform OCR.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence