Why PC-Based Machine Vision?
| By: Steve Mleczewski, Project Engineer
Why use a PC-based machine vision system when Smart cameras seem to dominate the vision world? While each technology has its place in the market PC-based systems may offer more than you think.
In the 1990s the progression of smaller chips and better circuit board technology gave us a cost point that allowed the birth of the Smart camera. Now manufacturers could put a processor, imager and I/O in one small package. All of a sudden road blocks such as size and cost were not factors on the factory floor. While the Smart camera revolution did greatly expand the vision market, it did not replace all PC-based systems.
PC-based vision systems generally consist of a PC with a Windows operating system. A frame grabber or Ethernet card in the PC interfaces to cameras to import the images. A software package on the PC analyzes the images, makes decisions and sends data such as pass/fail information to another system. Sound like 1987? That’s because it is. The hardware behind PC-based vision systems has changed very little over the last 20 years. The difference is that changes in computing power now allows for more powerful software. Following are some examples of what the computing power of PC-based vision can do.
PC-based vision can be more cost-effective and easier to use for multiple camera applications. Multiple camera applications include general inspection applications for finding the presence of parts. An example for this is inspecting for the presence of studs on the underbody of a car in an automotive body shop. An application like this may take 12 to 18 cameras. Using PC-based vision with all cameras linked back to the same interface is more cost-effective and easier to manage than using individual Smart cameras. This also holds true for bead inspection on large automotive parts. Inspecting the width and detecting gaps on adhesive beads usually requires multiple cameras to save inspection time.
Multiple cameras working together are also useful in 3D vision applications. Full 3D is defined as giving the X, Y, and Z location of a part as well as the rotations around X, Y and Z. To find the 3D position of large stamped parts or entire car bodies four cameras are generally required. Software on a PC-based system uses algorithms to find a relationship between features on the target to calculate the 3D position. The values for the 3D position are sent to a robot or PLC so that work may be done on the target.
Large CPU chip makers are always trying to make PC chips faster. Because of this PC-based vision systems have faster processors than Smart cameras. This allows PC-based vision to use high end algorithms for difficult applications such as OCR (Optical Character Recognition) in engraved metal applications. These types of applications can be tricky to light properly and usually have inconsistent characters. PC-based vision allows powerful tools to solve these types of applications much better than threshold-based tools.
The power of PC-based vision can also handle large resolutions. GigE and firewire cameras are available with resolutions of 5 megapixels and higher. Resolutions in this range are not available in Smart cameras.
A PC is required for programming and for visualization of most vision applications. With PC-based vision the end user always has access to the programming and debugging tools for the system. Password protection may be used to allow access to only certain individuals. Many PC-based vision systems also have visualization or HMI (human machine interfaces) to give feedback that most operators can understand. The image shows an example of an HMI with an image of a dashboard. The missing component shows up in red so that an operator can easily troubleshoot the problem without getting an engineer involved.
A PC can act as a gateway to communicate to almost any device. Vision systems usually have to communicate to a PLC or robot. A PC has communication options like serial and Ethernet built-in. In addition PCI cards for almost any standard industrial communication can be easily added to a PC. If the vision software doesn’t have a driver for a certain PCI card, technologies like OPC may be used to link them together.
More than just Cameras
Another advantage of PC-based vision is that the PC can interface to more than just cameras. Certain time-of-flight lasers can sweep across parts to create a 3D image that could not happen with standard CCD or CMOS cameras. An example application for this is bin-picking brake rotors. The image gives a 3D view of a bin full of rotors. This data is transformed to a color coded 2D image that is transferred to the PC-based vision software. The vision system then sends the best pick location to a robot.
Another application for this type of technology is reading DOT (Department of Transportation) codes on tires. The tire may be rotated around as a time-of-flight laser gathers data. The result is an image of the tire unwrapped so that OCR tools may be used to read the codes.
Both PC-based vision and Smart cameras have a place in the vision market. The next time you have a vision application you may want to consider the advantages of each technology. You may find that PC-based vision will do the job better.
VMT, part of the Pepperl+Fuchs Group, provides individual turnkey systems and complete solutions for automation of industrial image processing and laser sensor systems. The highly qualified VMT engineering team has more than 20 years of success in industrial image processing covering over 500 system installations.