New metrics target image sharpness in AI deblurring

Researchers have developed a new class of metrics called Sharpness Informed (SI) Image Quality Assessment (IQA) metrics, designed to specifically address sharpness in deblurred images. These metrics penalize over-sharpening, a common issue in deep neural network deblurring. A subjective study found that images restored using a sharpness-aware composite loss were preferred in 67% of comparisons over those without explicit sharpness targeting. The new SI-PSNR metric demonstrated superior performance compared to other PSNR variants on IQA benchmarking datasets. AI

IMPACT Introduces new evaluation metrics that could improve the quality of AI-generated images by better assessing and controlling sharpness.

RANK_REASON Research paper introducing new metrics and dataset. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• deep neural network
• SI-PSNR
• Uditangshu Aurangabadkar

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →