Integration and control
Industrial efficiency can be greatly enhanced through image processing. Vision systems replace a large number of sensors and are the basis for maximizing the automation of numerous processes.
Barcode reading, dimension capture, quality control, parts tracking and identification, and production line monitoring are just some of the possible applications. Automation and optimization result in greater efficiency, higher productivity and better quality, fewer revisions, and improved customer satisfaction.
An easy-to-integrate solution
The pervasive use of robots at an industrial level in many diverse sectors—from mechanics to electronics, and food to medicine—encourages the application of robots for a variety of purposes that often require specialized tools and technologies.
The artificial vision system is among the increasingly requested tools and is found in two versions: 2D and 3D—each with its own pros and cons.
A SIMPLIFIED BUT EFFECTIVE PLAN
OVERTECH assists its clients in choosing the correct vision technology and manages the configuration and fine-tuning.
We also offer advice and assistance on vision systems already being used by the client.
TWO-DIMENSIONAL ARTIFICIAL VISION
Enter the era of automation
In the case of 2D vision systems, the captured image of the target object is actually flat and only has two dimensions. The image does not provide any information relating to the height: there are data for the X and Y axes but not for the Z (depth). The actual image is an outline of a 3D object seen from a specific vantage point. Different observation points and different objects create completely different outlines, which makes the use of a 2D system very limited for functions where information regarding the shape is essential for performing an operation.
Modern functions
The lack of information regarding the actual shape and height is not a big problem for many purposes, so the 2D vision system—be it an area or line scanner—is widely used in the industrial sector for many tasks such as:
- verification of characteristics and position;
- size control;
- barcode reading;
- character recognition;
- label verification;
- quality control;
- presence detector.
In all these cases, the 2D image is obtained through variations of reflected light on the surface of the object being scanned. The contrast of the item being read—both in greyscale and in color—is the first challenge for 2D vision systems.
2D vision systems limitations
Light sensitivity: since the image of the object being scanned is obtained from light reflecting from it, ambient light variations can have a negative impact in terms of accuracy. Too much light, too little light, or shadows from the work environment can negatively affect the clarity of the edges and features that appear in the 2D image.
Absence of contrast: given that the 2D vision system depends on the contrast as defined on the surface of the object, the difficulty in reading very light or very dark objects must be taken into account.
Height-related errors: since no information on height is available, errors can occur due to the movement of the object on the Z-axis of the plane. If an object is always stationary on a perfectly flat surface, at a precise focal distance from the image sensor, image capture errors will be significantly reduced.
In all of the following situations, 2D vision systems fail to recognize the shape and perform the operation:
- with complex or assembled parts, where dimensions must be measured beyond the X and Y planes;
- when determining the volume of the object;
- when an object needs to be picked up and moved precisely.
THREE-DIMENSIONAL ARTIFICIAL VISION
Three-dimensionality
With a 3D vision system, the target object is no longer a flat figure, but a three-dimensional item composed of precise coordinates. The position of each pixel in the space is known, and simultaneously provides data for the X, Y, and Z axes.
Comparative interventions
There are 4 techniques for making a 3D vision system:
- laser triangulation;
- stereoscopic vision;
- trajectory time;
- structured light.
The Pick-it camera, for example, is based on stereoscopic vision.
Compared with two-dimensional image processing, the work in three dimensions requires more time and intense use of advanced processors and software—multi-core processors and 3D algorithms, for example—to manage the volume of a production line. However, by virtue of their ability to reliably capture the extra third dimension, 3D vision systems are not impacted by environmental factors that create difficulties for the 2D system. Aspects of brightness, contrast, and distance from the object are no longer a problem.
Real-life applications
Since the three-dimensional digitized model of the target object is more accurate, the robots can manage both shape and position. In fact, they know the precise position of the object in space, its exact volume, the surface of the angles, and the degrees of flatness, regardless of the light conditions in the work environment or whether the object is partially shiny or light-absorbing black.
As a result of this additional capability, the 3D vision system can be applied to a wide spectrum of uses where the characteristics of the 2D system are not sufficient:
- measurement of thickness, height, and volume;
- sizing and space management;
- measurement of shape, holes, angles, and curves;
- surface detection or assembly defects;
- quality control and verification with respect to 3D CAD models;
- robot orientation and surface tracing (i.e., for welding, gluing deburring, and more);
- container gripping for moving, packing, or assembling;
- scanner and object digitization.
