Pan Sharpening Techniques: Enhancing Multi-Spectral Image Resolution

Pan sharpening is a technique used to enhance the spatial resolution of multi-spectral (MS) images by incorporating high-resolution panchromatic (PAN) images. This research investigates various pan sharpening methods, including deep learning-based approaches, model-driven techniques, and neural network architectures, to improve the resolution and quality of MS imagery. The study evaluates the performance of different pan sharpening methods using metrics such as Peak Signal-to-Noise Ratio (PSNR), Enhanced Relative Global Accuracy Statistic (ERGAS), and Structural Similarity Index Measure (SSIM). The results indicate that the performance varies significantly across different datasets and methods, with PSNR values ranging from 24.7927 dB to 47.3063 dB, ERGAS values from 0.5425 to 6.7479, and SSIM values from $\mathbf{0. 9 0 7 5}$ to $\mathbf{0. 9 9 8 9}$. The analysis reveals that WorldView datasets are more commonly used in pan sharpening research, likely due to their high-quality and availability. This research contributes to the field of remote sensing by providing insights into the effectiveness of different pan sharpening techniques and their potential applications in various domains.