The goal of dynamic scene deblurring is to remove the motion blur present in a given image . We propose a novellow-level perceptual loss to make image sharper . To better focus on imageblurriness, we train a reblurring module amplifying the unremoved motion blur . Our experimental results show the proposed re-blurring losses improve the perceptual quality of the deblurred images in terms of NIQE and LPIPS scores as well as visualsharpness . The self-supervised loss at inference stage inspects if the debrurred image still contains noticeable blur to be amplified. Theself-supervisory reblurred loss at in-built tests show that the new loss function improves the image’s visual sharpness and NIQe and NPIPS scoreings are more effective than those of the previous loss function. The new loss loss function

Author(s) : Seungjun Nah, Sanghyun Son, Jaerin Lee, Kyoung Mu Lee

Links : PDF - Abstract

Code :
Coursera

Keywords : loss - image - function - blur - niqe -

Leave a Reply

Your email address will not be published. Required fields are marked *