MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641099609 A) filed by Dr. S. Senthamizh Selvi; Ms. R. Gayathiri; Ms. V. Radha; Ms. S. Deeparani; J. Lavanya; Ms. Nagamalini V.; Ms. Lavanya Jayaraman; and Ms. N. M. Sharmila on August 18, 2026, for Machine Vision Based Image And Video Analysis System For Industrial Defect Detection And Process Control.
Inventors include Dr. S. Senthamizh Selvi; Ms. R. Gayathiri; Ms. V. Radha; Ms. S. Deeparani; J. Lavanya; Ms. Nagamalini V.; Ms. Lavanya Jayaraman; and Ms. N. M. Sharmila.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: The present invention relates to an image and video analysis system based on machine vision for automatic industrial defect detection and process control. The system consists of: one or more industrial cameras, an illumination unit, an image and video acquisition module, an edge processing unit, an image preprocessing module, a feature extraction module, an artificial intelligence-based defect detection model, a defect classification unit, a process decision controller, an actuator control interface and a monitoring and data storage unit . During operation, images or continuous video frames of products, components, manufacturing surfaces, assemblies or process regions are recorded at predetermined inspection locations. The operations performed on the acquired visual data include noise reduction, contrast enhancement, geometric correction, illumination normalisation, region-of-interest extraction, image resizing and frame synchronisation. The processed images are fed into a machine-learning or deep-learning model for detecting visual and dimensional abnormalities. The artificial intelligence model identifies such defects as cracks, scratches, dents, holes, surface contamination, dimensional deviations, incorrect assembly, missing parts, colour variations, deformations, welding defects, coating irregularities and other specific manufacturing defects. The model detects defect location, defect type, severity level, confidence value and related product identity information. In video-based inspection, abnormal motion, process deviation, tool misalignment, improper material flow, machine vibration, assembly sequence errors and changes in manufacturing conditions are detected by analysing consecutive image frames. A process decision controller compares the detected defect or process abnormality with preset acceptance limits and generates an automatic process control command. Based on the condition sensed, the system can reject the defective product, stop or slow down the operation of the machine, change process parameters, raise an alarm, change actuator settings, divert a product for rework, or carry out further inspection. The control action is communicated to programmable logic controllers, industrial robots, motor drives, conveyor systems or other manufacturing equipment. After implementation of the control action the next images or video frames are analysed to determine if the process condition has returned to an acceptable state. The inspection results, defect images, process conditions, control actions and verification results are stored for traceability and model improvement. Real Machine Vision Based Image and Video Analysis System for Industrial Defect Detection and Process Control The proposed system therefore provides
Disclaimer: Curated by HT Syndication.
