A novel pansharpening method based on side window filter and new injection gain matrices

Area of Science:

• Remote Sensing
• Image Processing
• Geospatial Technology

Background:

• High-resolution remote sensing image acquisition is limited by spectral imaging technology.
• Pan-sharpening fuses multispectral (MS) and panchromatic (PAN) images to enhance spatial resolution.
• Existing methods often suffer from loss of image details during fusion.

Purpose of the Study:

• To develop a novel multispectral image fusion method for improved detail preservation.
• To introduce an enhanced adaptive Gram-Schmidt method based on Side Window Filtering (SWF).
• To address the limitations of traditional pan-sharpening techniques.

Main Methods:

• Application of Side Window Filtering (SWF) to PAN images for noise reduction and edge enhancement.
• Development of an improved adaptive Gram-Schmidt method (SWGSA) with an adjusted injected gain.
• Fusion of MS and PAN images using the SWGSA method for detail preservation.

Main Results:

• SWF effectively reduces noise and enhances edge information in PAN images, improving weight index computation.
• The adjusted injected gain, referencing the PAN image, further boosts image detail.
• Experimental validation on IKONOS, GeoEye-1, and WorldView-3 datasets confirmed significant improvements in fused image quality.

Conclusions:

• The proposed SWGSA method demonstrates superior performance in pan-sharpening compared to traditional techniques.
• The method effectively preserves and enhances image details, addressing a key limitation in remote sensing image fusion.
• The enhanced detail preservation leads to higher quality fused multispectral images suitable for various applications.