MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202621053847 A) filed by Dr. D. Y. Patil Institute Of Technology, Pimpri, Pune - on April 28, 2026, for Enhancing Visual Clarity In Satellite Imagery For Remote Sensing And Analysis.
Inventors include Dr. Rachna Somkunwar; Dr. Vinod Kimbahune; Ankit Pal; Gaurav Vibhute; Harshit Godawat; Kartik Gonde; and Sunita Patil.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: Image dehazing plays a vital role in enhancing image visibility by mitigating the effects of atmospheric scattering caused by fog, haze, and pollution. Traditional dehazing methods typically depend on simplified assumptions, such as uniform haze distribution, and require accurate estimation of physical parameters, which limits their performance in complex and dynamic real-world environments. Moreover, such approaches often involve high computational costs and may fail to fully recover fine image details under dense or non-uniform haze conditions. The present disclosure proposes a novel end-to-end deep learning architecture for image dehazing that addresses these challenges. The system is based on an encoder–decoder framework integrated with an enhanced multi-head attention transformer block, which utilizes wavelet transformation to capture multi-scale spatial and frequency features. This design enables improved representation of both global context and local textures. Additionally, a deformable gated attention mechanism is introduced within skip connections to facilitate adaptive feature alignment and selective information fusion across network layers, enhancing reconstruction quality. The proposed architecture eliminates the need for explicit haze parameter estimation by learning a direct mapping from degraded to clear images. It achieves superior dehazing performance with reduced computational complexity, making it suitable for real-time and resource-constrained applications across domains such as surveillance, autonomous systems, and remote sensing.
Disclaimer: Curated by HT Syndication.
