MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641072363 A) filed by Srm Trp Engineering College on June 11, 2026, for Spatial And Spectral Feature Integration For The Identification Of Morphed Facial Images.
Inventors include Keerthana M; Nasrin Begum S; Ramya S; Rashmika G K; and Dr. A. Muthuraj.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: Manipulated facial images, especially those created through basic morphing and editing tools, are increasingly being used for impersonation, misuse, and online harassment. Identifying such alterations is not straightforward, as many edits introduce only minor visual changes that are difficult to notice through manual inspection. Most existing solutions rely on computationally heavy models, which limits their usability on standard systems. This work presents a practical approach for examining facial images using a combination of texture and frequency-based analysis. The system studies local pixel relationships through Local Binary Pattern (LBP) to understand surface-level irregularities, while Fast Fourier Transform (FFT) is used to observe changes in frequency patterns that may indicate hidden modifications. By combining these two perspectives, the system is able to capture both visible and underlying signs of image tampering. The extracted information is evaluated using a simple comparison mechanism, where measured values are checked against predefined limits to determine the likelihood of manipulation. This avoids the need for model training while still providing consistent results. The design follows a clear sequence of steps, including preprocessing, feature extraction, evaluation, and result generation, making the overall process easy to follow and implement. The final output categorizes the image and presents the findings in a structured format, allowing users to understand the result without technical difficulty.
Disclaimer: Curated by HT Syndication.
