Attention Mechanism Enhanced Kernel Prediction Networks for Denoising of Burst Images

TL;DR

This paper introduces attention mechanism enhanced kernel prediction networks (AME-KPNs) that leverage cost-effective attention modules to improve burst image denoising by utilizing inter-frame and intra-frame redundancies.

Contribution

The paper proposes a novel AME-KPN architecture that refines feature maps with attention modules and predicts adaptive kernels, residuals, and weights for effective burst image denoising.

Findings

Demonstrates robustness of AME-KPNs in simulated and real-world scenarios

Achieves superior denoising performance compared to existing methods

Effectively utilizes redundancies within image bursts

Abstract

Deep learning based image denoising methods have been extensively investigated. In this paper, attention mechanism enhanced kernel prediction networks (AME-KPNs) are proposed for burst image denoising, in which, nearly cost-free attention modules are adopted to first refine the feature maps and to further make a full use of the inter-frame and intra-frame redundancies within the whole image burst. The proposed AME-KPNs output per-pixel spatially-adaptive kernels, residual maps and corresponding weight maps, in which, the predicted kernels roughly restore clean pixels at their corresponding locations via an adaptive convolution operation, and subsequently, residuals are weighted and summed to compensate the limited receptive field of predicted kernels. Simulations and real-world experiments are conducted to illustrate the robustness of the proposed AME-KPNs in burst image denoising.