SDGSAT-1, Thermal infrared spectrometer (TIS), Destriping, Maximum interclass variance (OTSU), Remote sensing, Physical geography, GB3-5030, Environmental sciences, GE1-350

Stripe noise is present in the on-orbit images captured by the Sustainable Development Goals Satellite-1 (SDGSAT-1) Thermal Infrared Spectrometer (TIS). Removing these stripes lays the crucial groundwork for subsequent remote sensing data applications. This paper proposes a high-performance method for removing stripe noise in SDGSAT-1 TIS large dynamic range imaging. Based on the fixation properties of the stripe noise, we utilize the a priori information to enhance the gradient variation around the stripe region. This step successfully separates the background from the noise and speeds up the noise extraction. We extract local thresholds from the central difference matrix using an adaptive threshold extraction method based on the Maximum Interclass Variance (OTSU). To maximize the retention of image details, we combine structural information about stripe positions with a weighted filter kernel based on the L2 norm to predict noise digital number (DN). The proposed method extensively experiments with simulated and on-orbit data in different scenarios. It has the best overall performance in qualitative evaluation, quantitative evaluation, processing speed, and robustness.