Physen-Noise2Noise: Physics-Guided Self-Supervised Defocus Deblurring with Bias Correction under Low-Light Conditions
arXiv:2605.24590v1 Announce Type: cross Abstract: Low-light, long-exposure defocus deblurring remains a challenging problem due to the simultaneous presence of severe blur and complex biased noise. Existing methods typically rely on simplified noise assumptions, which limits their effectiveness under realistic imaging conditions. In this work, we propose Physen-Noise2Noise, a self-supervised deblurring framework guided by the physical model of defocus imaging, which leverages noisy multi-frame observations without requiring clean reference images. Unlike conventional Noise2Noise-based approach
The proliferation of AI and advanced imaging capabilities is driving the need for more robust and reliable image processing techniques, especially under challenging conditions, pushing continued research in this domain.
Improving image quality in low-light and blurred conditions without requiring clean reference images has broad implications for computer vision applications, enhancing reliability and reducing data annotation costs.
This self-supervised method allows for more effective deblurring and noise correction in real-world scenarios, potentially expanding the operational envelope for various imaging systems.
- · Computer Vision Researchers
- · Surveillance and Security Industry
- · Autonomous Vehicle Developers
- · Computational Photography Companies
- · Companies relying on traditional, less robust deblurring techniques
Enhanced performance of AI models that rely on visual input under difficult lighting conditions.
Reduced sensor hardware requirements or extension of existing sensor lifespans in certain applications due to software improvements.
New applications become feasible in environments previously considered too challenging for automated visual analysis.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG