Convolutional Neural Networks for Direct Text Deblurring

article in a collection out of WoS and Scopus

English

In this work we address the problem of blind deconvolution and denoising. We focus on restoration of text documents and we show that this type of highly structured data can be successfully restored by a convolutional neural network. The networks are trained to reconstruct high-quality images directly from blurry inputs without assuming any specific blur and noise models. We demonstrate the performance of the convolutional networks on a large set of text documents and on a combination of realistic de-focus and camera shake blur kernels. On this artificial data, the convolutional networks significantly outperform existing blind deconvolution methods, including those optimized for text, in terms of image quality and OCR accuracy. In fact, the networks outperform even state-of-the-art non-blind methods for anything but the lowest noise levels. The approach is validated on real photos taken by various devices. Further information including test data and trained networks can be found on the [PROJECT PAGE] (http://www.fit.vutbr.cz/~ihradis/CNN-Deblur/).

convolutional neural networks, blind deconvolution, image restoration, deblurring, CNN, neural networks, deep learning

HRADIŠ, M.; KOTERA, J.; ZEMČÍK, P.; ŠROUBEK, F.

2015

16. 8. 2015

The British Machine Vision Association and Society for Pattern Recognition

Swansea

1-901725-53-7

Proceedings of BMVC 2015

1

13

13

http://www.bmva.org/bmvc/2015/papers/paper006/index.html