Adjustable model-based fusion method for multispectral and panchromatic images

Area of Science:

• Remote Sensing
• Image Processing
• Computer Vision

Background:

• Panchromatic (PAN) and multispectral (MS) image fusion aims to create high spatial resolution (HR) MS images.
• Existing methods often struggle to balance spatial enhancement with spectral information preservation.

Purpose of the Study:

• To develop an adjustable, model-based image fusion method for MS and PAN images.
• To improve the quality of fused HR MS images by adaptively controlling regularization parameters.

Main Methods:

• Formulated image observation models relating desired HR MS images to observed low-resolution MS and HR PAN images.
• Employed a maximum a posteriori (MAP) framework to solve the inverse problem of image fusion.
• Utilized a gradient descent algorithm with adaptive regularization parameter selection to balance spatial and spectral information.

Main Results:

• The proposed method was tested on QuickBird and IKONOS satellite images.
• Quantitative evaluation indices demonstrated superior performance compared to several existing fusion methods.
• Experimental results verified the method's efficacy in producing high-quality fused images.

Conclusions:

• The developed adjustable model-based fusion method effectively integrates spatial and spectral information.
• Adaptive parameter control is crucial for optimizing the trade-off between spatial enhancement and spectral fidelity.
• The method shows significant potential for applications requiring high-resolution multispectral imagery.