Gabor is Enough: Interpretable Deep Denoising with a Gabor Synthesis Dictionary Prior

TL;DR

This paper demonstrates that constraining a deep denoising neural network to use Gabor filters results in a highly interpretable model that achieves near state-of-the-art performance with fewer parameters, highlighting the sufficiency of Gabor-like representations.

Contribution

The authors introduce GDLNet, a denoising CNN constrained to learn only Gabor filters, showing that simple, interpretable filters can match complex models' performance.

Findings

GDLNet achieves near state-of-the-art denoising results.

The Gabor filter constraint reduces model complexity significantly.

The network maintains noise-level generalization capabilities.

Abstract

Image processing neural networks, natural and artificial, have a long history with orientation-selectivity, often described mathematically as Gabor filters. Gabor-like filters have been observed in the early layers of CNN classifiers and even throughout low-level image processing networks. In this work, we take this observation to the extreme and explicitly constrain the filters of a natural-image denoising CNN to be learned 2D real Gabor filters. Surprisingly, we find that the proposed network (GDLNet) can achieve near state-of-the-art denoising performance amongst popular fully convolutional neural networks, with only a fraction of the learned parameters. We further verify that this parameterization maintains the noise-level generalization (training vs. inference mismatch) characteristics of the base network, and investigate the contribution of individual Gabor filter parameters to the performance of the denoiser. We present positive findings for the interpretation of dictionary learning networks as performing accelerated sparse-coding via the importance of untied learned scale parameters between network layers. Our network's success suggests that representations used by low-level image processing CNNs can be as simple and interpretable as Gabor filterbanks.