Cutting-edge weld defect detection.
A cutting-edge platform to pinpoint critical weld defects. No guesswork, no delays!
Luminous software analyzes pipe images with welding defects by comparing them against an optimized reference database of ideal weld structures. Using advanced Machine Learning (ML) algorithms, it identifies and highlights deficiencies in real time.
Our Application
Our software automatically detects welding defects using radiation-based imaging, giving you fast, reliable results with the power of advanced computer technology.
Our Business
Luminous is an easy-to-use software solution offered by annual subscription, built on proven radiographic methods to deliver clear, reliable inspection results
Our Solution
“A non-destructive testing (NDT) solution designed for pipeline weld inspections, leveraging advanced computer technology to identify, interpret, and visualize defects in the welding area with precision and reliability.
Luminous Application Procedure
Luminous Canada Inc. offers powerful software that identifies issues in welded connections across all types of pipes and materials. Built for industries such as oil, gas, water, and food, it works in any weather and delivers reliable results wherever you need them.
Related Basic Technologies
Workflow
Subsequently, a set of geometrical features is extracted from the source as input to a classifier (CNN). Image segmentation is a commonly used technique partitioning an image into multiple segments or regions.
Pre-Processing
Pre-processing takes place through Machine Learning (Supervised Learning). The basic idea is to mimic the way a human inspector would inspect radioscopic images.
DWDI
Double Wall Double Image (DWDI) exposure technique is a typical arrangement adopted for taking radiographic images of the pipe with a diameter equal to or less than 80 mm, thereby not allowing any internal access for the insertion of the radiation source.
Pre-Trained DNNs
We make use of pre-trained DNNs to map the knowledge for Visual Recognition. As DNNs are machine learning mechanisms that comprise expanded Convolutional Neural Networks (CNNs or ConvNets), during feature extraction, image classification takes place through CNN or ConvNets.
Classification
These networks are typically applied to image classification, regression and feature learning, including prediction of series with Deep Long Short-Term Memory Neural Networks.
CNN
The CNN layer processes elementary visual features, such as edges and corners, located at different regions of the input. Once the match is made, the results can be viewed on a computer monitor remotely, or a mobile device.
Computer-Aided Radiographic Image Analysis for Weld Fault Detection
- Self-Sufficient
- Cost Management
- Expand Over Time
- Beyond Time Barriers
- Anyplace
- AI Technology
