Quality Magazine

Five Reasons to Invest in Vision Software

December 2, 2010
Several options are available that can be tailored to your vision system.

Vision is affordable for all kinds of different applications. Source: Cognex Corp.


Machine vision systems are devices that capture and analyze visual information and are used to automate tasks that require “seeing.” More specifically, machine vision systems are used to detect defects, monitor production lines, guide assembly robots, and track, sort and identify parts. Machine vision systems are programmed to perform narrowly defined tasks, which may include counting objects on a conveyor, reading serial numbers and searching for surface defects. They are particularly favored over simple sensors or manual systems for visual inspections that require high-speed, high-magnification, 24-hour operation and repeatability of measurements.

In the past, inspection of parts was done either by labor or by using very simple sensors to identify known defects or conditions. On the one hand, labor is much smarter than any vision system, but the quality of manual inspections is never consistent. Simple sensors often require the parts being inspected to be in a very precise position. This solution is very rigid because if new defects show up, more sensors are required.

With machine vision, camera lens and image sensors are used to capture images. Vision software, the core element of the computer vision system, is “the man behind the curtain,” analyzing what is being seen and communicating information to other equipment. Vision systems also require lighting and are configured by connecting them to a display device such as a computer monitor.

Operators can purchase vision software to use with their own camera and processor, or they can purchase a standalone unit that combines camera, processor and software into a single package.



VARIETY OF FLAVORS

The software required to run vision systems is about as varied as the uses to which it may be put. Vision software is available as a set of vision tools found in a C or C++ library, which is completely programmable to create a custom solution.

It also is available in an application configuration environment, where the operator creates the application by selecting and configuring vision tools. Here the operator configures the application flow and inputs/outputs and programming or scripting supports flexible solutions. Finally, vision software may be part of an application-specific solution, where the operator simply needs to adjust the existing vision functionality.



Vision software delivers part location and robot guidance, part identification, inspection and gaging, and dimensional measurement. Source: Cognex Corp.

ACQUISITION FLEXIBILITY

The number one reason for using vision software is the flexibility it allows for the source of the image acquisition. The software can be used for both frame grabbers and direct connect cameras. It can be used for high-speed devices (640 x 480 at 205 frames per second (fps) and 1,280 x 1,024 at 500 fps) and for high resolution devices, including 4,872 x 3,248 (16 megapixels) and 12,288 x 1 devices.

It also can be used for a wide variety of special imagers, including thermal imaging, three-dimensional height profiling and X-ray imagers.

  • ­Thermal imaging. Food inspection is one key use for thermal imaging, including measuring meat temperature and checking for bruises. In the automotive arena, thermal imaging is used for manufacturing quality control of such systems as window defrosters and car seat heaters. In the semiconductor market, thermal imaging systems pick up on hot spots or bad solder areas. Finally, medical imaging systems use thermal imaging for fever detection and fluid control.

  • Three-dimensional height profiling (laser stripe). This type of special imaging device is frequently used when features are not visible with a standard camera. Target applications may include shape and volume inspection for food, glue bead or solder balls; surface inspection to detect gross scratches or dents; tests for product integrity to determine the presence or absence of a feature, or its completeness; alignment to show co-planarity and part orientation; and finally, optical character recognition on stamped or embossed surfaces.

  • ­X-ray. X-ray devices are used for quality inspection to show the presence or absence of a feature, as well as integrity of the part in question. They also are used for volume measurement in quality control of bread, fruits and meat. X-ray imagers can be used for metal, glass, stones, bones and rubber, and also can show void or missing products. They cannot be used for paper, wood, nutshells, pits, foil or aluminum metals.

 



POWERFUL VISION TOOLS

Along with flexibility, vision software comes with an array of tools that can be brought to bear on an application, including tools for geometric pattern matching, finding parts in low contrast situations or when parts have flexible distortion, decoding, reading barcodes and verifying characters.

Color alignment has been on the market for some time, but new features that have the ability to tolerate rotation, scale and skew were previously only available on specialized tools. Objects can be located with synthetic geometries by using computer-aided design (CAD) drawings to train a model.
 
Robust character verification is possible, offering high confidence in rough environments. For defect inspection, the ability to tolerate rotation, scale and skew is a key requirement.


SCALABLE HARDWARE

Vision software can be run on hardware platforms scaled to the particular application, making them adaptable. For example, they can be run on ultra-small Intel Atom-based industrial PCs, which offer a relatively low price and an industrial design; desktop PCs, for the lowest price and highest performance; industrial rack PCs for maximum robustness; and finally, the smallest PC vision system can run on Atom-based smart cameras.


SIMPLE INTEGRATION

When things get complex, vision software offers a wide choice of I/O modules, including analog, digital, Modbus and Profibus. Third-party software can be used for faster integration, for example, HMI builder and Soft PLC. Multiple cameras, including Analog 4, FireWire 63, USB 127 and GigE 255.

For more complex applications, coding can be much more efficient than configuring the systems. Configurable systems have been designed to build simple applications. The benefit is a quick development of applications, even for beginners.
 
The downside of a configurable system is that some tasks are hard to describe. If, for example, one needs multiple retry strategies with a combination of partial results of each strategy, then coding would be a better way to proceed.
 
A good way to determine if a configurable solution has missed its sweet spot is to ascertain when an operator who needs to maintain a configured solution decides to start over from scratch because they do not understand what the originator was trying to do.


LOW HARDWARE COST

Last, but certainly not least, the overall hardware cost for a vision software-based system remains extremely low. The cost for the hardware needed to run a system are usually most economically viable for applications requiring multiple cameras and multiple installations of the same application. To these costs must be added the cost for the vision software and the cost for application development and deployment.

The primary business drivers for vision software are applications that are difficult to configure, those that require the highest flexibility and best performance, those that require multiple camera options or those that need the most customizable operator interface. Either force fit requirements into a configurable system with all its features and risk that compromises may have to be made on the application, or use the flexibility of vision software and tailor the solution to one’s needs-with some extra effort, of course. V&S


TECH TIPS

  • Vision software is an extremely powerful tool for part location and robot guidance, part identification, inspection and gaging, and dimensional measurement.

  • The number one reason for using vision software is the flexibility it allows for the source of the image acquisition.

  • Vision software can be run on hardware platforms scaled to the particular application, making them adaptable.