Improvement in panchromatic-guided denoising algorithm for research on Beijing SDGSAT-1 luminous remote sensing image denoising algorithm

The Glimmer Imaging Unit (GIU) onboard Sustainable Development Science Satellite 1 (SDGSAT-1) provides higher-resolution nighttime image data for the study of remote sensing images, but the noise in the images can have a negative impact on experimental research. In this study, based on existing denoising algorithms, the OTSU threshold method is combined with a convolution window to identify noisy pixels. The spectral correlation between the low-gain and high-gain panchromatic (PL and PH) and color (RGB) bands is used to remove noisy pixels and improve the quality of nighttime light images. Beijing is used as an example in this study. In the bands with RNE (residual noise entropy) and MRD (relative mean deviation) values close to 0, the denoising rate is maintained at about 80% in the full color and RGB bands. A comparison with the results of other commonly used denoising algorithms also shows its great advantages. The RNE value in each band is stable below 0.45, and the MRD value is 0. This study aimed to optimize the denoising process of nighttime remote sensing images through this method, providing clearer and more reliable data resources for research and applications in related fields.