Skip to content

hli1221/imagefusion_dllrr

Repository files navigation

Multi-focus Image Fusion using dictionary learning and Low-Rank Representation

Hui Li, Xiao-Jun Wu*

SpringerLink: https://link.springer.com/chapter/10.1007/978-3-319-71607-7_59

arXiv: https://arxiv.org/abs/1804.08355

ICIG2017(oral)

The framework for fusion method

Framework

Dictionary learning

Reconstructure

Figures and data

1 made_images and made_images_new are the source images which contain different focus region.

2 image_vector and image_vector_new are the image patches matrices and each column is an imape patch which divided by sliding window technique.

3 dictionary and dictionary_new are the su-dictionaries from image_vector and image_vector_new.

Source code

1 Hog.m---extract the HOG features of image patch.

2 The code of LRR

solve_lrr.m

solve_l1l2.m

inexact_alm_lrr_l1l2.m, inexact_alm_lrr_l1.m

exact_alm_lrr_l1l2.m, exact_alm_lrr_l1.m

3 getClassLabel.m ---- set class label for each patch.

4 fusion_dllrr.m ---- main file.

5 The tool boxes of KSVD and OMP

LRR parts

The LRR method is proposed by Guangcan Liu in 2010.

"Liu G, Lin Z, Yu Y. Robust Subspace Segmentation by Low-Rank Representation[C]// International Conference on Machine Learning. DBLP, 2010:663-670."

And we just use this method in our paper without change.

Citation

@inproceedings{li2017multi,
  title={Multi-focus image fusion using dictionary learning and low-rank representation},
  author={Li, Hui and Wu, Xiao-Jun},
  booktitle={International Conference on Image and Graphics},
  pages={675--686},
  year={2017},
  organization={Cham, Switzerland: Springer}
}

About

multi-focus image fusion using dictionary learning and low-rank representation(ICIG 2017, Shanghai)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages