Andrew Wilson, Editor, [email protected]
Using software standards, systems integrators can add value to their applications and spend less time developing low-level image-processing libraries.
Many automated machine-vision and image-processing systems control several processes based on the results of data captured from images. For example, robotic pick-and-place systems generally locate components and transfer them to the next process. To build such systems, systems integrators must understand both the intricacies of machine-vision technology and how they can be incorporated with existing robotic systems.
For high-speed image analysis, system integrators often must build systems that use high-speed data-acquisition devices and high-resolution cameras and frame grabbers. They must often seamlessly integrate disparate image-processing, data-acquisition, and input/output software.
Fortunately, software developers can turn to component-based object-oriented software such as the Common Object Request Broker Architecture (CORBA) from the Object Management Group (Needham, MA; www.omg.org) and the Component Object Model (COM) from Microsoft Corp. (Redmond, WA; www.microsoft.com). CORBA is a vendor-independent architecture and infrastructure that computer applications can use to work together over networks.
On Windows-based machines, COM is the basis of Microsoft's approach to distributed computing. A method for customizing applications, it forms the foundation of OLE (object linking and embedding) and ActiveX technologies. ActiveX offers extensible standards and mechanisms that enable software developers to package their functionality and content into reusable components. Whereas frameworks such as CORBA offer developers the means to develop custom and reusable image-processing software, their image-processing toolkits are limited to military, government, and university sites.
The same is not true for Microsoft's COM, where a number of software packages, including ActiveMIL from Matrox (Dorval, Quebec, Canada; www.matrox.com) and VisionPro COM/ActiveX software architecture from Cognex Corp. (Natick, MA; www.cognex.com), enable third-party components for process control, I/O, and machine control to be added to vision applications. Leading vendors of data-acquisition software are also selling ActiveX objects. These include other standards, such as OLE for Process Control (OPC) managed by the OPC Foundation (Scottsdale, AZ; www.opcfoundation.org), that define a set of standard interfaces based upon Microsoft's OLE/COM technology.
Traditionally, software or application developers needed to write a custom interface or server/driver to exchange data with hardware field devices. OPC eliminates this requirement by defining a common interface that permits this work to be done once and then allows reuse by interface, supervisory control and data acquisition control, and custom application packages. By applying the OPC standard interface, interoperability among automation/control applications, field systems and devices, and business applications is easily accomplished.
While OPC is designed to support factory-floor and process-control equipment, such as programmable logic controllers and distributed control systems, the Open Data Acquisition Standard from the Open Data Acquisition Association (Stamford, CT; www.opendaq.org) is tailored for data acquisition. Similar to the OPC standard, it also allows interoperability between PC-based data-acquisition hardware and software platforms among multiple vendors. However, the set of software objects also provides a software abstraction of the most common functions of PC-based boards, including analog input, analog output, digital input, digital output, and counter timers.
Using these software standards, systems integrators can add value to their specific applications and spend less time developing low-level image-processing libraries, graphical user interfaces, and data-acquisition or process-control software. Moreover, they can also shorten product time to market and reduce end-user product cost.
