MFIF: Attentive Multi-Focus Fusion Network (AMFFNet)

DOI

10.2991/978-94-6239-723-1_29How to use a DOI?

Keywords

Multi-focus images; Image fusion; CNN; Entropy; SSIM

Abstract

Multi-focus image fusion (MFIF) aims to synthesize the all-in-focus image from partially focused inputs, yet preserving fine details under non-uniform focus remains challenging. We present AMFFNet (Attentive Multi-Focus Fusion Network), a hybrid approach that combines frequency-domain analysis with deep learning. Inputs are decomposed via discrete wavelet transform (DWT) into multi-resolution sub-bands, passed through a shared CNN encoder, and refined by dual attention (channel and spatial) to emphasize informative cues. Features are then fused with an adaptive weighting strategy and reconstructed by inverse DWT to produce the final image. On the Lytro dataset, AMFFNet consistently surpasses classical and recent CNN-based methods on SSIM, Entropy, Q_AB/F, and VIF, and yields sharper, more structurally coherent results in visual comparisons. The architecture preserves detail and generalizes well to real-world focus variations, offering a practical solution for high-quality MFIF.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Bharat Bhardwaj
AU  - Ajit Kumar Jain
PY  - 2026
DA  - 2026/07/14
TI  - MFIF: Attentive Multi-Focus Fusion Network (AMFFNet)
BT  - Proceedings of the International Conference on Responsible, Risk-aware, and Regulated AI (RRRAI 2026)
PB  - Atlantis Press
SP  - 314
EP  - 323
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-723-1_29
DO  - 10.2991/978-94-6239-723-1_29
ID  - Bhardwaj2026
ER  -