MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541122779 A) filed by B V Raju Institute Of Technology, Narsapur, Telangana, G06T, on Dec. 5, 2025, for 'a novel multi-path generator gan approach for realworld haze reconstruction and removal.'
Inventor(s) include Nagaram Ramesh; Rowthu Neelima; Kenchali Shivaji Rao; J Manikandan; Sriharsha Vikruthi; V Indumathi; Mastan Mohammed Meera Durga; and Boda Sunitha.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "This invention discloses a novel Multi-Path Generator Generative Adversarial Network (MPG-GAN) specifically developed for reconstructing and removing haze from real-world images. Unlike conventional single-stream dehazing frameworks, the proposed system introduces a uniquely structured three-branch generator architecture, consisting of a structural analysis path, a color-restoration path, and a depth-approximation path. Each of these branches is designed to capture a different dimension of haze characteristics, enabling the model to learn heterogeneous haze patterns that typically arise under varying environmental, lighting, and atmospheric conditions. The structural path focuses on preserving edges, contours, and fine-grained textures that are often degraded by haze. The color-restoration path compensates for chromatic distortions, ensuring natural and visually consistent color reproduction. The depth-approximation path estimates scene depth cues, which are critical for accurately removing spatially varying haze densities. Outputs from these three specialized paths are then adaptively fused to generate high-quality, haze-free images that retain both perceptual realism and structural integrity. To further enhance the reliability of the system, a hybrid discriminator framework is incorporated, employing multi-scale adversarial learning to enforce perceptual fidelity, structural consistency, and artifact suppression. This comprehensive adversarial strategy enables the network to differentiate subtle haze artifacts from genuine scene details. The proposed MPG-GAN offers significant advantages over existing dehazing methods, including superior reconstruction accuracy, improved visual clarity, and robust performance across diverse and unpredictable real-world scenarios. By eliminating the dependence on handcrafted physical priors or rigid atmospheric models, the invention provides a flexible, data-driven solution capable of delivering advanced haze-free image restoration in practical applications."
Disclaimer: Curated by HT Syndication.
