ICIF: Image fusion via information clustering and image features

Area of Science:

• Computer Vision
• Image Processing
• Sensor Fusion

Background:

• Current image fusion techniques struggle with poorly correlated features from different sensors, leading to reduced clarity.
• Existing fusion strategies often fail to adequately compensate for inter-sensor feature discrepancies.

Purpose of the Study:

• To propose a novel image fusion method, Information Clustering and Image Features (ICIF), to overcome limitations of existing techniques.
• To enhance the correlation between image features and improve the clarity of fused images.

Main Methods:

• Utilizing a weighted median filter algorithm for image clustering, incorporating infrared texture features to guide visible light image clustering.
• Decomposing images into base, bright detail, and dark detail layers to improve inter-layer correlations.
• Employing sensor-derived characteristics and inter-layer features as weights for fusion strategy, reconstructing images based on extended texture details.

Main Results:

• The proposed ICIF method demonstrates superior performance compared to state-of-the-art methods on public datasets.
• ICIF effectively highlights targets and preserves abundant image details.
• The method's generalizability was shown by successful application to medical and multi-focus image fusion.

Conclusions:

• The ICIF method offers a significant advancement in image fusion technology, improving clarity and detail preservation.
• ICIF provides a robust framework for fusing images from diverse sensor types, with broad applicability.