The Devil is in the Upsampling: Architectural Decisions Made Simpler for Denoising with Deep Image Prior

by   Yilin Liu, et al.
University of North Carolina at Chapel Hill

Deep Image Prior (DIP) shows that some network architectures naturally bias towards smooth images and resist noises, a phenomenon known as spectral bias. Image denoising is an immediate application of this property. Although DIP has removed the requirement of large training sets, it still presents two practical challenges for denoising: architectural design and noise-fitting, which are often intertwined. Existing methods mostly handcraft or search for the architecture from a large design space, due to the lack of understanding on how the architectural choice corresponds to the image. In this study, we analyze from a frequency perspective to demonstrate that the unlearnt upsampling is the main driving force behind the denoising phenomenon in DIP. This finding then leads to strategies for estimating a suitable architecture for every image without a laborious search. Extensive experiments show that the estimated architectures denoise and preserve the textural details better than current methods with up to 95 makes them especially robust to a higher level of noise.


page 1

page 5

page 7

page 8


Zero-shot Blind Image Denoising via Implicit Neural Representations

Recent denoising algorithms based on the "blind-spot" strategy show impr...

Practical Blind Denoising via Swin-Conv-UNet and Data Synthesis

While recent years have witnessed a dramatic upsurge of exploiting deep ...

Details Preserving Deep Collaborative Filtering-Based Method for Image Denoising

In spite of the improvements achieved by the several denoising algorithm...

Rethinking Deep Image Prior for Denoising

Deep image prior (DIP) serves as a good inductive bias for diverse inver...

Image Denoising for Strong Gaussian Noises With Specialized CNNs for Different Frequency Components

In machine learning approach to image denoising a network is trained to ...

Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators

Convolutional Neural Networks (CNNs) have emerged as highly successful t...

Please sign up or login with your details

Forgot password? Click here to reset