Image Fusion Method Based on Snake Visual Imaging Mechanism and PCNN

Area of Science:

• Computer Vision
• Biomimetic Engineering

Background:

• Image fusion enriches image quality for better analysis, with infrared and visible image fusion gaining importance.
• Deep learning is prevalent in image fusion, but often requires large datasets, which are not always available.
• Snake visual mechanisms, processing both infrared and visible information, offer a bionic approach to image fusion without extensive training data.

Purpose of the Study:

• To develop a novel infrared and visible image fusion method inspired by snake visual mechanisms.
• To address the limitation of unclear details in existing snake-inspired fusion methods by integrating a pulse-coupled neural network (PCNN).
• To create a fusion technique that does not necessitate large amounts of training data.

Main Methods:

• Investigated receptive field models of retinal nerve cells and dual-mode cell imaging mechanisms of rattlesnakes.
• Developed mathematical models based on snake visual systems and integrated them with a pulse-coupled neural network (PCNN).
• Proposed an improved fusion algorithm for infrared and visible images based on these bionic principles.

Main Results:

• The proposed fusion method was tested on eleven sets of source images.
• Performance was evaluated using three non-reference image quality assessment indices.
• The improved algorithm demonstrated superior overall performance compared to seven other fusion methods across all three indices.

Conclusions:

• The bionic-inspired, PCNN-enhanced image fusion method effectively improves image quality and detail.
• This approach overcomes the need for large training datasets, making it suitable for data-scarce applications.
• The proposed method represents a significant advancement in infrared and visible image fusion technology.