MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202641071894 A) filed by Malla Reddy Engineering College For Women Autonomous; Malla Reddy University; Malla Reddy Mr Deemed To Be University; and Malla Reddy Vishwavidyapeeth Deemed To Be on June 10, 2026, for System And Method For Detecting Artificial Images Using Local Binary Pattern.
Inventors include Dr. Y. Madhaveelatha; Ms Manasa Spandana Donkini; Ms Sudheshna Peruri; Mr Putta Kishorekumar; Mr. Gundabattina Madhavarao; Dr. G. Gifta Jerith; Dr. Pattlola Srinivas; and Mr. Chekuri Mahesh.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: The current invention reveals a state-of-the-art Autonomous Data Cleaning Agent which is especially developed to solve the signal degradation phenomenon of noisy sensor environment. In industrial IoT (IoT) and remote monitoring, sensors are often exposed to strenuous environments that add the noise of different types, including high frequency electromagnetic interferences with sensors, sensor drift and baseline wandering. Traditional methods of filtering mostly use fixed parameters that cannot manage the dynamic nature of these disturbances resulting in either under filtering where noise is retained or over filtering where useful signal data may be lost. The invention removes these constraints by putting an intelligent agent on the job, which is able to diagnose the particular type of noise profile in real-time and actively choosing the most suitable remediation strategy. The system is designed as a multi-layered system that lies between the downstream analytics platform and the layer of the raw sensor data acquisition. This agent does not use predefined filters like that used in traditional hard-coded filters but uses a lightweight inference engine to examine the statistical properties of the incoming data stream, e.g., variance, kurtosis, and spectral density. On the basis of such an analysis, the agent separates the real signal anomalies, which are real world physical phenomena, and those that arise due to environmental noise or system failure. This difference is very important in applications like predictive maintenance which may be used to respond to a sudden spike, which could indicate machinery failure instead of merely being a sensor spike. Lastly, agent autonomy minimizes the amount of operational overhead required to operate large scale sensor networks. With a traditional configuration, data engineers have to manually tune the filters on various sensors, which is incompatible with networks of thousands of nodes. This calibration is automated by the current invention making it easy to deploy and scale. The outcome is an efficient, self-sustaining data pipeline that guarantees the high-fidelity inputs to critical decision- making procedures in such industries as manufacturing and agriculture as well as smart city infrastructure. The present invention relates to a system and method for detecting artificially generated images using texture-based feature extraction techniques. With the rapid advancement of generative artificial intelligence models such as GANs and diffusion-based systems, distinguishing between real and synthetic images has become increasingly challenging. The proposed invention utilizes the Local Binary Pattern (LBP) algorithm to analyze micro-texture variations in digital images. The system performs preprocessing operations including image resizing, grayscale conversion, and normalization, followed by extraction of texture descriptors using the LBP technique. These descriptors are converted into feature vectors and processed through machine learning classification algorithms to determine whether an image is real or artificially generated. The invention provides a lightweight and computationally efficient detection mechanism suitable for real-time deployment in digital forensic analysis, cybersecurity systems, and social media verification platforms.
Disclaimer: Curated by HT Syndication.
