Focused CNN Framework: An Inspector Ensemble Approach with Patch-Based Voting for Multi-Focus Image Fusion

DOI

10.2991/978-94-6239-723-1_28How to use a DOI?

Keywords

Multi-focus image fusion; Ensemble learning; Convolutional networks; Patch processing; Decision mapping

Abstract

Multi-focus image fusion creates one clear image from multiple input images taken at different focal settings. Traditional techniques typically rely on manually crafted sharpness measures or frequency transforms, often producing blocking artifacts and struggling with complex textures. This paper presents a patch-based design combining lightweight convolutional “inspectors” in an ensemble, resolving decisions through confidence-weighted voting and patch-aware blending. Each inspector predicts which source patch is sharper and provides a reliability score used for aggregation, consistency checking, and spatial refinement. The design focuses on explainability using clear decision maps, with efficiency at approximately 0.87 million parameters and speed approximately 2.3 times faster than reconstruction networks. Evaluated on three multi-focus benchmarks—Lytro, MFFW, MFI-WHU—using PSNR/SSIM, QAB/F, mutual information, and runtime metrics, results demonstrate strong edge preservation, robustness to minor registration errors, and a favorable speed–accuracy trade-off via patch size adjustment.

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  - Aman Sharma
AU  - Prajjwal Singh
AU  - Aryan Bansal
AU  - Arpit Verma
PY  - 2026
DA  - 2026/07/14
TI  - Focused CNN Framework: An Inspector Ensemble Approach with Patch-Based Voting for Multi-Focus Image Fusion
BT  - Proceedings of the International Conference on Responsible, Risk-aware, and Regulated AI (RRRAI 2026)
PB  - Atlantis Press
SP  - 301
EP  - 313
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-723-1_28
DO  - 10.2991/978-94-6239-723-1_28
ID  - Bhardwaj2026
ER  -