Improving OCR Accuracy on Images with Motion Blur via GAN Derivatives
Kyoungwan Woo · 2020
This research aims to remove motion blur from images to increase conversion accuracy of OCR (optical character recognition) based programs, by utilizing generative adversarial network derivatives. Utilizing a self-created dataset, images were taken with and without motion blur to be used in the training of the network. The final images were fed into OCR programs to compare their accuracy.