Towards Controllable Real Image Denoising with Camera Parameters

2025-07-02

AI Summary

Existing deep learning denoising methods lack dynamic control over denoising strength conditioned on noise level, camera parameters (e.g., ISO, shutter speed, f-number), and user preferences. To address this, we propose the first physically grounded, controllable image denoising framework that explicitly incorporates camera parameters: these parameters are encoded into a learnable control vector and embedded into a deep denoising network, enabling joint optimization of data-driven representations and physics-based imaging priors. This design endows the model with explicit, interpretable, and user-controllable denoising strength adjustment. Extensive experiments demonstrate that our method can be seamlessly integrated—“plug-and-play”—into mainstream denoising architectures, yielding significant performance gains on both synthetic and real-world images. Notably, it exhibits superior robustness and generalization under complex noise distributions and varying camera configurations.

Technology Category

Computer Vision: Low Level & Physics-based VisionSearch and Optimization: Learning to SearchMachine Learning: Neuro-Symbolic Learning

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingUser Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendationResponsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Recent deep learning-based image denoising methods have shown impressive performance; however, many lack the flexibility to adjust the denoising strength based on the noise levels, camera settings, and user preferences. In this paper, we introduce a new controllable denoising framework that adaptively removes noise from images by utilizing information from camera parameters. Specifically, we focus on ISO, shutter speed, and F-number, which are closely related to noise levels. We convert these selected parameters into a vector to control and enhance the performance of the denoising network. Experimental results show that our method seamlessly adds controllability to standard denoising neural networks and improves their performance. Code is available at https://github.com/OBAKSA/CPADNet.

Problem

• Adjust denoising strength based on noise levels
• Utilize camera parameters for adaptive noise removal
• Enhance denoising performance with ISO, shutter speed, F-number

Innovation

• Utilizes camera parameters for denoising control
• Converts ISO, shutter speed, F-number into vector
• Enhances standard denoising network performance