BM3D vs 2-Layer ONN

TL;DR

This paper compares simple two-layer neural networks, including self-organized ONNs, with BM3D for image denoising, showing that Self-ONNs outperform CNNs and can be competitive or superior to BM3D, especially at high noise levels.

Contribution

It demonstrates that compact Self-ONNs can outperform CNNs and rival BM3D in image denoising, highlighting their efficiency and effectiveness in resource-constrained scenarios.

Findings

Self-ONNs outperform CNNs in denoising tasks.

Self-ONNs are competitive with BM3D, especially at high noise levels.

Two-layer networks can achieve significant denoising performance.

Abstract

Despite their recent success on image denoising, the need for deep and complex architectures still hinders the practical usage of CNNs. Older but computationally more efficient methods such as BM3D remain a popular choice, especially in resource-constrained scenarios. In this study, we aim to find out whether compact neural networks can learn to produce competitive results as compared to BM3D for AWGN image denoising. To this end, we configure networks with only two hidden layers and employ different neuron models and layer widths for comparing the performance with BM3D across different AWGN noise levels. Our results conclusively show that the recently proposed self-organized variant of operational neural networks based on a generative neuron model (Self-ONNs) is not only a better choice as compared to CNNs, but also provide competitive results as compared to BM3D and even significantly surpass it for high noise levels.