Sparsity and Autocorrelation Peak Difference Based Joint Noise Level and Blur Extent Estimation
Journal
Proceedings - 2025 21st IEEE Asia Pacific Conference on Circuits and Systems, APCCAS 2025
Start Page
1
End Page
5
ISBN (of the container)
979-833158907-3
Date Issued
2025-10-12
Author(s)
Abstract
The levels of blur and noise play a crucial role in guiding adaptive image restoration and enhancement. This is particularly important due to the inherent trade-off between denoising and deblurring: denoising may oversmooth fine details in sharp regions, while deblurring can introduce artifacts in smooth or noisy areas. To address this challenge, we propose a two-stage method for jointly estimating the noise level and the blur extent of an image. In the first stage, we estimate the noise level and perform denoising to reduce its interference with subsequent processing. In the second stage, we assess the global blur extent from the denoised image. In this work, an efficient and content-invariant noise estimation technique enhanced by sparsity-based compensation and a reliable blur quantification strategy that utilizes difference of peaks (DoP) are proposed. With these techniques, both the noise level and the blur extent can be estimated in an accurate way.
Event(s)
2025 21st IEEE Asia Pacific Conference on Circuits and Systems, APCCAS 2025
Subjects
adaptive restoration
blind quality assessment
blur estimation
image degradation analysis
noise level estimation
Publisher
IEEE
Type
conference paper