Multi-Focus Image Fusion Method for Vision Sensor Systems via Dictionary Learning with Guided Filter.

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Vision sensor systems (VSS) generate numerous images in surveillance, traffic, and industrial applications.
• Limitations in vision sensors often result in images lacking complete focus, hindering analysis.
• Effective multi-focus image fusion is crucial for improving image understanding in various contexts.

Purpose of the Study:

• To propose a novel multi-focus image fusion method, termed SRGF (Sparse Representation Guided Fusion).
• To address the challenge of obtaining all-focused images from vision sensor systems.
• To enhance the clarity and analytical value of images acquired by VSS.

Main Methods:

• Utilizes sparse coding to classify focused and defocused regions, generating focus feature maps.
• Employs a guided filter (GF) for calculating score maps and refining decision maps.
• Incorporates consistency verification to improve the accuracy of the final fused image.

Main Results:

• The proposed SRGF method achieves satisfying fusion results.
• Experimental validation demonstrates the effectiveness of the developed fusion technique.
• The method shows competitive performance compared to existing state-of-the-art fusion approaches.

Conclusions:

• The SRGF method provides a robust solution for multi-focus image fusion.
• This technique enhances image analysis capabilities in applications relying on VSS.
• The proposed approach is a valuable contribution to the field of image fusion technology.