Attention-Guided Matching for Stereo Multi-Focus Image Fusion

Sicheng Jiang; Xiaodong Chen; Huaiyu Cai; Yi Wang

doi:10.3788/AOS250836

Acta Optica Sinica, Volume. 45, Issue 15, 1510002(2025)

Attention-Guided Matching for Stereo Multi-Focus Image Fusion

Sicheng Jiang, Xiaodong Chen^*, Huaiyu Cai, and Yi Wang

Key Laboratory of Optoelectronics Information Technology, Ministry of Education, School of Precision Instrument and Optoelectronics Engineering, Tianjin University, Tianjin 300072, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Framework of proposed method. (a) Feature extraction; (b) feature matching; (c) feature fusion; (d) feature reconstruction

Download full size

Fig. 2. Architecture of encoder/decoder components. (a) CNN encoder; (b) Transformer encoder; (c) decoder

Download full size

Fig. 3. Workflow of modality alignment using Q-Former

Download full size

Fig. 4. Architecture of match module

Download full size

Fig. 5. Architecture of fusion module

Download full size

Fig. 6. Experimental dataset samples. (a) Original clear monocular images; (b) monocular near-focused images; (c) monocular far-focused images; (d) stereo multi-focus images, left view; (e) stereo multi-focus images, right view; (f) random masks

Download full size

Fig. 7. Comparative results of monocular multi-focus image fusion

Download full size

Fig. 8. Comparative results of stereo multi-focus image fusion

Download full size

Table 1. Quantitative results of monocular multi-focus image fusion

View table

Table 1. Quantitative results of monocular multi-focus image fusion

Metrics	NSCT-SR	GF	DSIFT	MFCNN	IFCNN	Ours
AG	4.8373	4.8453	4.8635	4.8720	4.8522	4.9151
GLD	9.4455	9.5174	9.6732	9.4765	9.8336	10.1928
NMI	0.8463	0.8715	0.9583	0.9172	0.9356	1.1034
VIFF	0.9539	0.9455	0.9548	0.9523	0.9806	0.9773
Q^AB/F	0.7318	0.6854	0.7209	0.7663	0.7689	0.7894

Table 2. Quantitative results of monocular multi-focus image fusion

View table

Table 2. Quantitative results of monocular multi-focus image fusion

Metrics	NSCT-SR	NSCT-SR*	DSIFT	IFCNN	IFCNN*	Ours
AG	6.6582	6.7029	6.7815	7.1801	7.2335	7.5258
GLD	11.0439	11.1553	11.2516	11.3777	11.5162	11.8949
NMI	0.8791	0.9065	0.8968	0.9336	1.0658	1.2104
VIFF	0.8304	0.8664	0.8590	0.8979	0.9644	0.9813
N^AB/F	0.1837	0.1531	0.1472	0.1581	0.1287	0.0715

Table 3. Ablation study results

View table

Table 3. Ablation study results

Metrics	Baseline		w/o match		w/o fusion
Metrics	Mono	Stereo	Mono	Stereo	Mono	Stereo
NMI	1.1034	1.2104	1.0975	0.9652	1.0217	0.9853
VIFF	0.9773	0.9813	0.9782	0.9284	0.9633	0.9576
Q^AB/F	0.7894	—	0.7867	—	0.7624	—
N^AB/F	—	0.0715	—	0.1512	—	0.1025

Tools

Get Citation

Copy Citation Text

Sicheng Jiang, Xiaodong Chen, Huaiyu Cai, Yi Wang. Attention-Guided Matching for Stereo Multi-Focus Image Fusion[J]. Acta Optica Sinica, 2025, 45(15): 1510002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Apr. 3, 2025

Accepted: Apr. 27, 2025

Published Online: Aug. 15, 2025

The Author Email: Xiaodong Chen (xdchen@tju.edu.cn)

DOI:10.3788/AOS250836

CSTR:32393.14.AOS250836

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology