Machine vision systems are used in a wide variety of applications, but the most prevalent use is the inspection of parts during the manufacturing process. The technology is used for assembly verification, dimensional measurements, parts identification, flaw detection, character verification or reading, robot guidance, process monitoring, and trend analysis. Machine vision enables companies to improve productivity while reducing production costs and the number of defective products shipped. (Click here for the sidebar "How Does Machine Vision Work?".)

With the rapidly increasing use of machine vision, many types of systems have been developed. Here is a look at vision systems manufactured as complete packages. These consist of specialized processing hardware and dedicated vision software. But just as important as the system itself are the lighting and optics (click here for the sidebar "Lighting and Optics").

Types of Vision Systems
Vision systems come in three basic configurations, classified by the type of processing performed: binary, gray, or color. Binary systems perform the most basic visual inspections and are low cost and generally easy to use. They operate on an image made up of black and white pixels and calculate results by counting the pixels to determine the shape or size of an object.

Binary systems allow inspection of silhouettes, profiles, and outlines and are best suited for area measurement and sizing. Applications such as verification of the size and length of capacitors and of parts placement and size in potentiometer subassemblies are ideal for binary processing. On the other hand, this type of system performs poorly in applications involving low-contrast or variable light, and they cannot detect imperfections in product appearance.

Gray scale imaging typically processes images using 256 levels of black and white. A popular method of processing gray scale images compares the incoming image with a template, or model, image. Product acceptance or rejection is determined by the degree of similarity between the two images-a technique referred to as correlation.

Gray scale processing provides more accurate dimensional measurements than binary systems because of a technique called subpixel processing, which is the ability to define an edge to an accuracy of less than one pixel. With this resolution, surface imperfections, scratches, textures, shadows, and shades of gray can be detected.

But gray scale processing does have drawbacks. It costs more than binary systems, and in some instances, binary processing is faster than gray scale processing because less data must be processed. Still, gray scale processing systems are a better choice than binary systems for applications involving low-contrast or variable lighting.

Gray scale or binary processing is sufficient for most industrial inspections, but for some applications, 256 gray levels are still not enough. What is needed is color imaging. For example, color is crucial in determining whether food is blemished or whether the correct pill was placed in a blister pack. Color processing uses information from the red, blue, and green color spectra to detect and differentiate shades of color relevant in such applications.

When compared with binary and gray scale processing systems, color systems are expensive. Higher computational requirements for the image data and the cost of the color camera drive up the price of the system.

Configurations
Because temperature, humidity, dust, and water can be harmful to a vision system, you must consider the factory environment when selecting a system configuration. Three configurations are available: stand-alone, PC-based, and VME-based. All three offer some degree of modularity for customization.

Stand-alone systems are "black boxes" that offer features that make them well suited for factory automation. The systems use ASICs that deliver accelerated processing power, making them among the fastest systems available. In addition, many stand-alone systems work seamlessly with other factory automation devices (e.g., PLCs, photoelectric sensors, and RF/ID systems), which provides greater integration of plant floor production and data tracking. Finally, these rugged systems withstand high temperatures, vibration, and electrical interference; this makes it possible to use them near solenoid valves and other inductive loads and motors without disrupting operations.

A PC-based system generally consists of a computer and a processing board with dedicated vision ASICs. Use of the computer's CPU is also becoming a popular option because it significantly lowers the price of the vision system. Using the CPU also allows you to take advantage of advances in microprocessor technology and declining prices.

However, PCs lack the robustness of stand- alone systems. Most PC components are designed for desktop use, not for harsh factory environments. Rugged computers and components are available, but their high cost often negates the original cost advantages of basic PCs.

VME-based systems combine the advantages of a robust system with a flexible user platform. VME is an IEEE bus standard that has proven reliability in industrial inspection applications. System components are connected via a pin-and-socket connection to the bus, improving signal transfer and enduring vibrations better than the edge connector used on PC peripherals.
The major disadvantage of these systems is their high cost. In addition, VME-based vision processors and peripherals are complex and cannot ride on the coattails of advances made in the PC marketplace.

Programming Methods
You can configure many vision systems without doing any programming. These turnkey systems meet the requirements of a single application. They offer quick setup but provide only limited flexibility.

Other systems use a menu interface, which integrates an internal program with a point-and-click selection mechanism. This method is easy to use and offers a moderate degree of flexibility. Visual Basic and C++ have emerged as the most popular custom programming languages for vision systems, but many systems use proprietary programming languages. To implement this type of system, you must be a proficient programmer and conversant in vision algorithms. Systems integrators specializing in vision are often used to develop vision programs. Typically, the integrator will also set up the system on the plant floor, design the lighting scheme, and install the fixturing.

What Else Should I Know?
A well-researched specification will help you identify the most appropriate system for your application and avoid paying for more than what is required. In developing the specification, you should answer the following questions:

• How many parts must be inspected per second or minute? Processing speeds vary from system to system.
• Which features distinguish a good part from a bad part? Any part variations-such as size, color, texture, and positional or rotational deviations-should also be identified. When speaking with a systems integrator, it is helpful to show both good and bad sample parts.
• What environmental factors-such as light variations (e.g., sunlight through a window), dust, and oil-are present in your factory? This will determine how rugged your system must be. Many of these factors also affect lighting.
• What level of technical knowledge will the system operator possess? This helps determine the best type of user interface.
• Who will install, program, maintain, and integrate the system? Do you have in-house expertise, or will you work with a systems integrator? The human and monetary resources allocated to the project will help determine the type of system you choose.
• Will any other machines or components be interfaced with the vision system? If so, how far do the signals have to travel? This information is necessary to determine the communications configuration and which results should be sent to which machine.
• What are your upgrade and flexibility requirements? Any significant changes in part characteristics or throughput, for example, will affect system operation. You should consider whether parts added to your production line will differ substantially from existing parts. If your system is not flexible enough, you may have to purchase another vision system for the new application.
• What is your budget?

Home | Contact Us | Advertise © 2009 Questex Media Group, Inc.. All rights reserved. Reproduction in whole or in part is prohibited. Please send any technical comments or questions to our webmaster.