MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641050075 A) filed by Ambika Annavarapu; and Surekha Borra, Guntur, Andhra Pradesh, on April 20, for 'system and method for adaptive deep learning-based denoising of multi-mission remote sensing imagery.'
Inventor(s) include Ambika Annavarapu; and Surekha Borra.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to a sensor-adaptive deep learning system for denoising multi-mission remote sensing data acquired from diverse spaceborne and airborne imaging platforms. Remote sensing data captured across multiple missions-such as optical, multispectral, hyperspectral, and radar systems-are affected by heterogeneous noise patterns arising from sensor characteristics, radiation exposure, atmospheric interference, and platform-specific electronics. Conventional denoising approaches fail to generalize across missions due to fixed noise assumptions. The proposed system introduces a sensor-adaptive framework that dynamically identifies sensor metadata and mission-specific noise characteristics, and accordingly selects or configures a deep learning denoising model optimized for the detected sensor domain. The system integrates spectral-spatial attention mechanisms and adaptive noise estimation to provide robust, cross-mission performance. By enabling automatic adaptation across multiple remote sensing platforms, the invention enhances image quality, improves downstream analytics, and reduces manual recalibration efforts. The proposed method supports scalable deployment in ground-based processing systems as well as edge-enabled on-board platforms."
Disclaimer: Curated by HT Syndication.
