MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641125895 A) filed by Karpagam Academy Of Higher Education; Karpagam College Of Engineering; Mrs. S. Gayatri; and Sridhar M on October 22, 2026, for Sar Image Colorization.
Inventors include Mrs. S. Gayatri; and Sridhar M.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Synthetic Aperture Radar (SAR) imaging plays a crucial role in remote sensing applications due to its capability to acquire high-resolution images under all weather conditions and during both day and night. Unlike optical imaging systems, SAR sensors utilize microwave signals that penetrate clouds, fog, smoke, and precipitation, making them indispensable for continuous Earth observation. However, SAR images are typically represented in grayscale and are affected by speckle noise and complex backscattering effects, which significantly limit their visual interpretability. This lack of intuitive visual cues makes SAR image analysis challenging, especially for non-expert users, and often requires extensive domain knowledge for accurate interpretation. This project presents an artificial intelligence–based approach for the automated colorization of SAR images to enhance their visual clarity and semantic understanding. The proposed system leverages deep learning techniques to transform monochromatic SAR imagery into realistic and visually meaningful color representations while preserving essential structural and textural information. By converting grayscale radar images into colorized outputs, the system aims to bridge the gap between complex SAR data and human visual perception, thereby facilitating improved analysis and faster decision-making. The core methodology employs a combination of Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs). CNNs are utilized for effective feature extraction, capturing spatial patterns, edges, and texture characteristics inherent in SAR images. The GAN framework consists of a generator network responsible for producing colorized images and a discriminator network that evaluates the realism of the generated outputs by distinguishing them from reference images. Through adversarial training, the model learns accurate color distributions and contextual relationships corresponding to various land-cover types and surface features. To address SAR-specific challenges, the system incorporates preprocessing techniques such as speckle noise reduction and intensity normalization, which improve model robustness and training stability. The colorization process is designed to maintain the geometric integrity of SAR imagery while enhancing visual interpretability. The proposed solution operates in a fully automated manner, eliminating the need for manual color mapping or expert intervention and enabling scalable processing of large SAR datasets. The effectiveness of the proposed approach is demonstrated through qualitative and quantitative evaluations, showing improved visual quality and enhanced interpretability of colorized SAR images. The system has broad applicability across multiple domains, including environmental monitoring, urban planning, disaster management, agriculture assessment, and defense surveillance. By enhancing the accessibility and usability of SAR imagery, this project contributes to the advancement of AI-driven remote sensing and offers a novel solution for intelligent SAR image visualization and analysis.
Disclaimer: Curated by HT Syndication.
