Gurpreet Kaur ¹ Manisha Malhotra ¹ Dilbag Singh ² Sunita Singhal ^3*

• ¹ Chandigarh University, Mohali, Punjab, India
• ² Lovely Professional University, Phagwara, Punjab, India
• ³ Manipal University Jaipur, Jaipur, Rajasthan, India

Pansharpening is an important remote sensing task that attempts to produce high-resolution multispectral (MS) images by combining low-resolution MS images with high-resolution panchromatic (PAN) images. Though deep learning-based pansharpening has shown impressive results, majority of these models frequently struggle to balance spatial and spectral information, resulting in artifacts and a loss of detail in pansharpened images. Furthermore, these models may fail to properly integrate spatial and spectral information, resulting in poor performance in complex scenarios. These models also encounter issues including gradient vanishing and overfitting. Therefore, this paper proposes a dual-path and multi-scale pansharpening network (DMPNet). It consists three modules such as the feature extraction module (FEM), the multi-scale adaptive attention fusion module (MSAAF), and the image reconstruction module (IRM). The FEM is designed using two paths such as primary and secondary path. The primary path captures global spatial and spectral information using dilated convolutions. The secondary path focuses on fine-grained details using shallow convolutions and attention-guided feature extraction. The MSAAF module adaptively combines spatial and spectral data across different scales, employing a self-calibrated attention (SCA) mechanism for dynamic weighting of local and global contexts and a spectral alignment network (SAN) to assure spectral consistency. Finally, to achieve optimal spatial and spectral reconstruction, the IRM decomposes the fused features into low-and high-frequency components using discrete wavelet transform (DWT). Extensive experimental results and evaluations reveal that the DMPNet is more efficient and robust than competing pansharpening models.