The Mechanical Eye: Machine Vision Technology

The Artificial Eye: Machine Vision Technology
By Jenny Bieksha, Bishop & Associates Inc.

Vision technology relates to the technology of artificial vision. Many job skills are based on visual perception—the ability to see, recognize, and distinguish objects, and to estimate distances. Vision technology deals with images or sequences of images, with the objective of manipulating and analyzing them in order to improve image quality, restore images, code pictures, or interpret images. Vision technology combines lighting, optics, electronics, information technology, software, and automation technology, and may be applied wherever images are generated and need to be analyzed.

Many of today’s production technologies extend beyond the limits of human visual capacities. This is where machine vision technology comes in. Cameras and computers give machines or devices the ability to “see,” to recognize objects or situations, and based on accumulated data, to make the correct decisions. Machine vision refers to the industrial application of vision technology and describes the understanding and interpretation of technically obtained images for controlling production processes. It has evolved into one of the primary technologies in industrial automation and is increasingly being used in many non-industrial applications

A Diversity of Systems
Machine vision systems are multifaceted. They check quality, guide machines, control processes, identify components, read codes, and supply valuable data for use in optimizing production. Industrial production accounts for the largest segment of machine vision systems. Two key components to maintaining manufacturing competitiveness in the Western economy are automation and quality. Good vision practices can help reduce costs and ensure that quality is consistent and meets high standards, thus increasing competitiveness.

Machine vision systems identify defective components at the earliest possible stage of production. Post-production costs are avoided when machine vision systems are used to enable a 100% inspection. Not only are defects identified, they are avoided. The electronic eyes of machine vision constantly monitor complex processes at all critical stages, identify changes in measured values, and signal the need for countermeasures so that manufacturing errors are ruled out. Control over complex production processes is increased, while the risk of costly downtime is reduced.

Some machine vision systems are so compact you can hold them in the palm of your hand; others could fill a whole room. Complete systems can be housed in increasingly smaller casings. Smart cameras and vision sensors can be integrated in places where little space is available. Machine vision systems may be classified into several categories, however, the actual systems and their applications are quite diverse.

Application-specific vision systems are turnkey systems for a specific application, such as the inspection of flat glass or wafers. They are usually PC-based and high performance.
Configurable vision systems are usually PC-based. Unlike application-specific systems, they are more versatile in their use. End users can often implement a range of applications via a graphical user interface (GUI).
Smart cameras are a complete vision system that includes software for image analysis, all contained within the compact housing of the camera body. They can be programmed flexibly to carry out different types of tasks.
Vision sensors are also a complete vision system within a single compact housing. In contrast to smart cameras, they are made for one specific type of application, i.e. code reading.

Vision Systems
In a vision system with multiple smart cameras, each camera performs single-point, local vision tasks, and then feeds the results over the network to a central processor or to a central PC. Most of these smart cameras must be configured by a PC, and then operated either by another PC or by a central processor. Embedded vision systems, sometimes called vision appliances or vision controllers, offer an alternative to traditional smart cameras and PC-based systems. The PC-based vision system provides basic connectivity in a manufacturing enterprise; it transmits data and images to plant machinery and connects to databases and SCADA systems. Embedded vision systems are used in scientific contexts, such as failure analysis labs and R&D, as well as in industrial applications.

Smart Cameras

Embedded vision systems have a processor for image processing and multiple I/O lines, but they also connect to external cameras for multi-camera inspection, and support a greater range of image sensors. In contrast, a smart camera is a camera that integrates a processor running an RTOS (real-time operating system) with an image sensor. It is a self-contained, standalone vision system with a built-in image sensor in the housing of an industrial video camera. Smart cameras typically consist of an image sensor, a processor, a communication interface, a lens holder or built-in lens, a built-in illumination device, a real-time operating system, and a video output.

Smart cameras ease system integration tasks, and high-speed, infrared, and X-ray imaging technologies increase the number of applications for machine vision systems. Containing a dedicated processor in each unit, smart cameras are especially suited for applications where several cameras must operate independently and often asynchronously, or when distributed vision is required (i.e. multiple inspection or surveillance points along a production line).

Vision Sensors
Vision sensors were introduced to bridge the gap between photoelectric sensors and complete machine vision systems. A standard photoelectric sensor allows a user to confirm that an object is in the inspection zone, but a vision sensor confirms the presence of the right object. Sensors typically have dedicated functionality or more limited functionality than vision systems, so they can be easier to use because the operator has fewer decisions to make.

A stand-alone vision sensor can replace many of the functions of a smart camera, barcode reader, or photoelectric sensor. It is usually small and can be integrated into tight spaces. It is not a vision system. However, as these sensors get smarter and their capabilities increase, the line between them and vision systems is blurring. Vision sensors are useful for applications requiring measurement or recognition. The trend toward ease of use and a broader range of applications on the factory floor will continue to grow. The gap between vision sensors and vision systems will shrink considerably.

Applications
In addition to high utilization in production environments, machine vision systems and components are used in a wide variety of applications:

High speed part inspection
Part sorting and identification
3D robotic guidance and material handling
Ultrasonic welding
Web inspection (inspection of continuously flowing materials, such as coils, tubes, wires, and extruded plastic) for defect detection and dimensional gauging
Medical robots for surgery
Unattended surveillance (detection of intruders, fire, or smoke)
Biometric recognition and access control (face, fingerprint, or iris recognition)
Defense and security, including space and satellite support, unmanned drones and land-based robots
Mapping underwater marine wildlife populations
Service robots in domestic applications, such as cleaning
Humanoid robots in space to perform dangerous extra-vehicular activity

Connectivity

There is no single digital interface that works best for all vision applications. Machine vision systems are enabled by image capture components such as GigE, Camera Link, FireWire, and USB cameras, CCD and CMOS image sensors, cables, and extenders. Sales figures for 2010 indicate that GigE Vision is now the most widely used interface in machine-vision cameras, exceeding those sold with a FireWire (IEEE 1394a/b) interface.

Healthy Forecast for the Vision Technology Market
Global competition will continue to intensify. Factories will be designed to yield maximum benefit from the lowest possible consumption of material and energy. It will be normal to have the strictest quality specifications in place. This transformation requires a pace of innovation more rapid than ever before. Machine vision will play a key role in this because it offers solutions suited to meet the challenges of the future. Vision system products will contain more horsepower and more pixels, so they can handle more data, more accurately. The integration of machine vision into machinery will continue, up to where the technology becomes a core part of the machine and is no longer an add-on to it.

Sales in 2010 of machine vision components and systems in North America alone were estimated at $1.8 billion, according to new figures released by the Automated Imaging Association (AIA). Sales growth was recorded in all major machine vision supplier markets, including: cameras, lighting, optics, imaging boards, application-specific machine vision systems, software, and smart cameras. Machine visions industry analysts anticipate healthy growth in 2011, contributed by built-up demand, new installations, and new markets—occurring mainly in government, security, medical, environmental technologies, and consumer applications.

Director, Renewable Energy, Medical, and Test, Measurement, and Instrumentation, Bishop & Associates Inc.
Jenny Bieksha joined Bishop & Associates in 2008 as its market segment director for the renewable energy, and the test, measurement, and instrumentation markets. She is currently a management consultant specializing in strategic business planning, with an emphasis on the development of program, market, and product plans. Bieksha has more than 20 years of experience in the electronics industry, with a background in market management, business development, channel sales, product management, and operations for ITT Corporation, Delphi Connection Systems, and Hughes Aircraft Company.

Bieksha has a bachelor of science degree in marketing from the University of Wyoming, and has since received her certificate as a project management professional.