High-quality image multi-focus fusion to address ringing and blurring artifacts without loss of information

For many multi-focus image fusion methods, fused image visual and quantitative analysis is a challenging problem because of the presence of ringing and blurring artifacts. A high-quality image fusion framework to overcome the challenges of ringing and blurring artifacts in multi-focus image fusion w...

Full description

Saved in:
Bibliographic Details
Published in:The Visual computer 2022-12, Vol.38 (12), p.4353-4371
Main Authors: Jagtap, Nalini Santosh, Thepade, Sudeep D
Format: Article
Language:English
Subjects:
Artificial Intelligence
Blurring
Computer Graphics
Computer Science
Computer vision
Decomposition
Deep learning
Edge detection
Fuzzy logic
Image filters
Image Processing and Computer Vision
Image quality
Morphology
Original Article
Wavelet transforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:For many multi-focus image fusion methods, fused image visual and quantitative analysis is a challenging problem because of the presence of ringing and blurring artifacts. A high-quality image fusion framework to overcome the challenges of ringing and blurring artifacts in multi-focus image fusion while preserving the input image regions in the fused image proposed. The modified guided filtering proposed is called novel information preservation-based guided filtering (IPGF). Each input image has been decomposed, using IPGF, into a detail image (small-scale details) and base image (large-scale intensity variations) using the edge detection operator (EDO). The EDO has been used to generate the guidance image to overcome the loss of detail parts and smooth tiny details. Low-rank representation (LRR) was used to estimate the focus map and generate the detail parts fusion. The inherent property of removing artifacts using LRR helps to generate artifact-free detail image fusion. A simple but effective choose max-based fusion has been applied to fuse base images. The addition of base and detail images leads to the initial fusion result. The guided filtering was applied to a focus map, and then the outcome was used to generate the final fused image. Guided filtering is applied to suppress the ringing and blurring effects from the final fused image. The proposed method shows improved performance in terms of statistical, subjective, and objective analysis compared to recent methods.
ISSN:0178-2789
1432-2315
DOI:10.1007/s00371-021-02300-5