INDUSTRIAL COMPUTER VISION
Information and automatic process control in production lines
Computer vision systems for factories
Industrial artificial vision provides information to factory systems so they can make decisions or perform processes automatically.
We develop solutions based on advanced computer vision from the design stage to its factory implementation. We adapt to the needs and requirements of our customers by customizing the development of each solution. We work from systems designed adhoc using open source libraries and neural networks to systems based on commercial solutions.
We defend the interests of our customers in each project thanks to our experience in various languages / architectures and our extensive network of computer vision technology manufacturers.
What do we do in computer vision?
Algorithm development or adaptation of commercial systems
We work with a wide variety of languages and environments to provide our customers the system that best suits their needs. OpenCV Python, C ++, SCIKIT-Learn, Python, Lab View, Halcon, Omron software…
Communication and implementation
Data visualization, communications and connection of the designed systems with pre-existing facilities in our customers factories. We can design and modify any system that we develop, so that it adapts to the mode of use desired by our customers.
Deep Learning
Neural networks and learning algorithms developed by us and adapted to the needs of our customers. We also use software and algorithms already trained and released for use. Depending on the application, we adapt the necessary development and complexity of the algorithms that we use.
Hardware
We use a wide variety of computer vision hardware, always thinking of the most suitable for the application. We work with a wide network of suppliers that supply us from smart cameras to industrial PCs, specific illumination, etc.
Technologies associated with computer vision
We design the complete vision system, both hardware and software.
The sensor or camera is the heart and the most important element of a computer vision system, in ATRIA we work mainly with 4 technologies and our solutions combine computer vision and deep learning algorithms and neural networks to generate a robust system.
2D Vision
With the 2D vision we can implement applications for character recognition, metrology or code reading, among others. It is the simplest sensor but it gives solution to many applications.
Thermography
Thermographic images allow to detect and visualize a temperature image of the area to be inspected. It is very useful in detecting leaks or load imbalance.
Hyperspectral Vision
The interaction of the light of different spectra with matter allows obtaining information about the composition and its properties. It is especially used in quality control.
3D Vision
3D vision is used in parts picking or defect detection applications. There are also different 3D reconstruction technologies that allow us to adapt to each application.
What applications does computer vision have?
Defect inspection
Metrology and fault detection
Quality checks
Identification of parts or intruders
3D reconstruction
Verification of the assembly of parts
Code reading and optical character recognition (OCR)
Vision + robotics for complex pickings
Data visualizations
Our computer vision projects
Code Reader / Optical Character Recognition (OCR)
Information from an alphanumeric code can be obtained through images. In addition, it is possible to perform character reading, very useful for digitizing information or improving traceability in production processes. We use our own systems with artificial intelligence that adapt to each specific case, which makes them much more flexible.
Advanced pattern recognition
Objects, shapes or pieces of many types can be located in an image. Previously, an algorithm capable of recognizing them in any of the possible positions (rotated, rotated, moved, etc.) is trained. It is very useful for locating complex forms. It is used in location, tracking or intrusion detection applications.
Metrology and fault detection
With computer vision, very precise measurement applications can be implemented. There are specific cameras with micron accuracy at great distances from the object. This way it is possible to measure online and automatically. The cycle times of these processes are usually below the second. For each application we select the optics, camera, illumination and computing unit necessary to meet the requirements of precision, repeatability and cycle time required.
3D reconstruction
3D cameras allow, through different technologies, the reconstruction of objects in three dimensions. It allows measuring distances between different planes of the same piece or calculating depths. It is used in the detection of defects, burrs, cracks, pitting …
Robotics + Computer vision
The cameras give ‘intelligence’ to the robotic arms. In this way, you can automate changing processes, or with a lot of variability. They adapt to the modifications and are able to react autonomously to them.
