Multi-Focus Image Fusion Algorithm Based on Supervised Learning for Fully Convolutional Neural Networks

Heng Li; Liming Zhang; Meirong Jiang; Yulong Li

doi:10.3788/LOP57.081015

Laser & Optoelectronics Progress, Volume. 57, Issue 8, 081015(2020)

Multi-Focus Image Fusion Algorithm Based on Supervised Learning for Fully Convolutional Neural Networks

Heng Li^1、*, Liming Zhang^2,3、**, Meirong Jiang², and Yulong Li¹

Author Affiliations

¹School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China

²Faculty of Geomatics, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China

³National-Local Joint Engineering Research Center of Technologies and Applications for National Geographic State Monitoring, Lanzhou, Gansu 730070, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Framework of proposed method

Download full size

Fig. 2. Network structure diagram

Download full size

Fig. 3. Constructed data

Download full size

Fig. 4. Training results of the 1st, 25th, 50th, 75th, and 100th epochs

Download full size

Fig. 5. Neural network loss function curve of training 100 epochs

Download full size

Fig. 6. Partial of multi-focus test images used in experiment

Download full size

Fig. 7. Experimental results of first group images with different algorithms

Download full size

Fig. 8. Experimental results of second group images with different algorithms

Download full size

Table 1. Average values of quality metrics for 27 test images under different fuzzy settings

View table

Table 1. Average values of quality metrics for 27 test images under different fuzzy settings

Index	SSIM	PSNR	CC	UQI
P2	0.8663	54.0889	0.9753	0.7928
P4	0.8842	57.1438	0.9793	0.8194
P8	0.8842	57.6086	0.9793	0.8195
P16	0.8842	57.6086	0.9793	0.8195
C2	0.8631	53.5615	0.9743	0.7863
C4	0.8813	56.8458	0.9788	0.8136
C8	0.8837	57.6728	0.9795	0.8200
C16	0.8837	57.6728	0.9795	0.8200

Table 2. Average values of quality metrics for 27 test images
View table
Table 2. Average values of quality metrics for 27 test images
Method SSIM PSNR CC UQI
NSCT 0.8758 57.1025 0.9778 0.8160
DTCWT 0.8724 56.9115 0.9769 0.8069
CNN 0.8742 56.9447 0.9765 0.8155
NSST 0.8806 57.9811 0.9793 0.8175
Proposed 0.8837 57.6728 0.9795 0.8200

Tools

Get Citation

Copy Citation Text

Heng Li, Liming Zhang, Meirong Jiang, Yulong Li. Multi-Focus Image Fusion Algorithm Based on Supervised Learning for Fully Convolutional Neural Networks[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081015

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites