A Subjective Study on a New Sharpness Informed Class of Metrics

2026-08-14

AI Summary

This study addresses the limitations of perceptual losses in DNN-based deblurring, specifically their lack of explicit sharpness optimization and absence of over-sharpening penalties. To overcome these issues, we construct a uniform sharpness dataset and propose a novel SI-IQA metric alongside a sharpness-aware composite loss function that explicitly penalizes over-sharpening. Extensive four-protocol subjective evaluations demonstrate that the proposed SI-PSNR metric achieves superior correlation with human perception. Furthermore, models trained with the new loss are preferred in 67% of comparative experiments. This work effectively bridges the gap in sharpness control within existing methods, significantly enhancing both the visual quality and perceptual consistency of deblurred images.

Abstract

Perceptual loss functions in Deep Neural Network (DNN) deblurring architectures improve the overall quality of restored images. However, few focus on explicitly targeting sharpness in the restorations. We conduct a subjective study of models trained with and without losses which explicitly target sharpness using a four-protocol approach, exploring preferred sharpness levels and effects on image quality. We introduce a novel dataset of images with uniform sharpness increments along with Difference Mean Opinion Scores (DMOS). Additionally, we propose a novel class of Sharpness Informed (SI) Image Quality Assessment (IQA) metrics which properly penalize over-sharpening. Our new SI-PSNR metric outperforms all other PSNR variants in terms of correlation statistics on IQA benchmarking datasets. We show that, on average, images restored using a sharpness-aware composite loss are preferred in 67% of binarized comparisons, as opposed to losses that do not explicitly target sharpness.

Problem

• Image Deblurring
• Perceptual Loss
• Sharpness
• Image Quality Assessment
• Over-sharpening

Innovation

• Sharpness Informed Metrics
• SI-PSNR
• Perceptual Loss
• Image Quality Assessment
• Subjective Study