Machine Vision Looks Well Beyond Inspection

Depth and diversity of applications show machine vision's strength beyond its dominant use for inspection. Also, overall purchases are expanding. About 35% of respondents expect machine vision purchases to increase over the next year, with about 54% expecting about the same amount spending, according to research by Control Engineering/Reed Research Group.

By Mark T. Hoske May 1, 2005

Depth and diversity of applications show machine vision’s strength beyond its dominant use for inspection. Also, overall purchases are expanding.

About 35% of respondents expect machine vision purchases to increase over the next year, with about 54% expecting about the same amount spending, according to research by Control Engineering /Reed Research Group. Survey results are based on responses from 182 subscribers, all of whom buy, specify, or recommend machine vision products.Machine vision use is in double digits for more than 15 applications; see graphic.Top applications are inspection, bar-code reading, motion control, gauging, and robotics. Largest growth from 2004 to 2005 occurred in motion control; continuous processing; verification; diagnostics, testing, maintenance; SCADA; machine control; CNC equipment; and discrete product manufacturing.

Among those using vision, 61% do so for in-plant requirements; 17% for OEM (original equipment manufacturer) resale requirements; and 22% for both OEM and in-plant requirements.

Application intelligence

‘Applications involving the pick and place of automobile door panels and hoods are becoming an increasing requirement for the automation solution suppliers to the automotive industry,’ according to William J. Amato, president, Phoenix Automation Group Inc. ( www.phoenixautomationgroup.com ).

‘This would involve the application of sensing and guidance technologies for the true three-dimensional guidance of industrial robotic systems,’ Amato says. ‘While previously this application has been accomplished with sophisticated ‘high-end’ vision and laser-based systems, we are discovering that the low-cost and ease of installation of DVT SmartImage Sensors…have worked in many such applications.’ Applications include sorting of automotive body styles; measurement or gauging of precision machined components such as fasteners, transmissions, and other subassemblies; as well as error-proofing for manufacturing, Amato adds.

Robotics applications have special needs, says Mark Sippel, Cognex In-Sight vision sensors principal product marketing manager. ‘The increased use of vision for guidance of a motion control system requires a very fast vision system that can synchronize image capture, processing, and analysis with the motion controller. Accuracy is another key to success in these applications because they require some kind of scaling factor that relates pixels to distances, and the resolution of the vision system correlates with the accuracy and repeatability of the motion system,’ Sippel says.

Via Ethernet

Regarding networks attached to machine vision, in 2004, survey respondents predicted that their use of Ethernet would soon overtake RS-232 as the dominant network. Ethernet did move from 65% in 2004 to 77% in 2005, and RS-232 moved from 73% down to 69%. Other notable increases in networking use with vision systems, 2004 to 2005, were DeviceNet from 36% to 40%, Modbus from 22% to 28%, and Profibus-DP from 20% to 25%.

Joshua Jelonek, machine vision application engineer, Keyence Corp. of America, concurs, ‘Ethernet is definitely the future of control communications. Its speed and flexibility offers many advantages to engineers looking to collect data for analysis or traceability purposes. It should be a priority for machine vision suppliers to provide their customers with powerful, easy-to-use data acquisition software to meet this increased demand.’ Other emerging network communication methods, Jelonek says, include EtherNet/IP, Modbus/IP, Profinet, and OPC.

About 52% of those responding used smart vision sensors in 2005, compared to 46% in 2004. About 83% said those smart visions sensors met requirements, compared to 90% in 2004. Write-in comments on satisfaction mentioned superiority of smart sensors compared to human inspection, ease of programming, and high success rates. Write-in reasons for dissatisfaction mentioned concerns about accuracy.

Kyle Voosen, machine vision product manager, National Instruments, says, ‘It seems to me that the big statistic here is that smart cameras’ meeting customer requirements is down from 90% to 83%. This probably has a lot to do with customer expectations on what smart cameras are capable of doing. It may also be a symptom of low cost. With smart cameras selling for such low prices, one can afford to risk a few thousand dollars just to see if a smart camera will work. Any concern about accuracy is a software issue. For all this talk about rugged, small vision systems, perhaps customers are more frustrated by the simple configuration software that comes with smart cameras.’

Performance, support, ease

In the survey results, the leading three product selection criteria when choosing machine vision are performance, technical support, and ease of use. See graphic. Seven requirements ranked higher than price in importance.

Dan Holste, director of vision products for Banner Engineering, observes, ‘Many of the product-selection criteria cited in the survey are about ease of use, critical when operators with multiple duties support highly sophisticated vision products on the factory floor. Performance is number one on the list of product selection criteria, and it should be. The sensor has to be able to do the job or it’s a no show, despite low cost and ease of use.

Continues Sippel from Cognex, ‘This year’s significant decrease in vision systems meeting user requirements may result from engineers specifying lower cost vision platform implementations that lack the hardware and software performance to adequately solve user applications. Perhaps this is one reason why demand for technical support, full product tool sets and more integrator expertise has increased, while the importance of product price has decreased.’

Perceptions of cost didn’t change much: 57% of respondents say machine vision isn’t too complex or costly to implement; 43% say it is, about the same as last year. Not surprisingly, in 2005, capital budgets are less of an impediment to increasing machine vision investments. In 2004, capital was the key obstacle to investing in machine vision for 46%. Just 39% saw that as the key problem in 2005. Priority compared to other automation was an impediment for 14% in 2004 and for 19% in 2005.

Other obstacles in the 2005 survey to increased use of machine vision were: 12% understanding of vision technology; 10% acceptance by factory personnel; 10% engineering resources; 4% says it’s difficult to use, and 7% had other reasons. Holste, from Banner, says he thinks ‘Lack of acceptance is driven both by ease of use—understanding the vision product and its use—and fear of job loss. Smart companies address both issues so they can be more competitive and produce higher quality products.’

Adaptation continues

System integrators are getting more machine vision business, perhaps to help those with lack of in-house resources. Integrators used for vision system projects jumped from 19% in 2004 to 25% in 2005; those planning to use increased slightly, 16% to 18%.

Increased spending on integration and on vision systems may suggest the market’s pleased, according to Robert Lee, vision product marketing manager, Omron Electronics. Expansion ‘is specific to the nature, progression, and development of vision technology products, focusing on its simplicity, product rich features and high success rates,’ he says.

Market evalution continues after more than 30 years. Ben Dawson, director of strategic development, ‘ipd,’ a division of Coreco Imaging Inc., says, ‘Machine vision was introduced in the early 1970s and found its first major market in electronics inspection and assembly starting in the 1980s. Factors in this success include the precise nature of manufactured electronic parts, the highly controlled environment for inspection, and willingness of electrical and process engineers to struggle with the technology. OCR and bar-code scanning evolved in the same period, but were driven by different market forces. In the last 15 years vision vendors have pushed machine vision technology to inspect parts with more variability, in messy environments, and have made a major effort to make their products easy-to-use.’

Continuing improvement in vision software algorithms will expand vision into areas of higher variability, such as food processing, Dawson adds.

Machine vision products

For more manufacturers, visit www.controleng.com/buyersguide . For integrators with vision expertise, go to www.controleng.com/integrators . For more on products, visit the Websites listed below.

Wafer reader; software upgrade

DVT Corp. enters the semiconductor world with a new wafer reader introduction; it also recently upgraded its free vision software.

The wafer reader is said to provide product traceability and real-time validation manufacturers require throughout the semiconductor fabrication process. Integrated lenses and lighting provide flexibility to read code consistently in production stages that change wafer appearance, such as contrast and color modifications. Acquisition of MTI Machine Vision in 2004 combines a ‘world-class algorithm library’ with DVT smart camera technology to create ‘the most robust OCR and decoding product offering in the vision industry,’ DVT says; which is ‘highly-reliable’ at a ‘significantly lower price point than is currently available.’ Introductory pricing is $4,995.

Free DVT machine vision software, Intellect version 1.1, has new algorithms with application-specific tools for bead tracking, fixed-font reading, feature count, and dynamic scaling. User interface improvements include ‘dockable windows,’ a default-restore command, and improved ‘snap to shape’ drawing tools. ‘System Explorer Window’ allows the user to browse global fonts and background scripts; and an enhanced ‘Properties Window’ with hierarchical tree control gives a streamlined, application-dedicated parameter layout. Support includes free classroom, online and CD-based product training, free operating software and upgrades, and free online diagnostics and troubleshooting. www.dvtsensors.com DVT

2-megapixel vision sensor

In-Sight 5403, a 1,600 x 1,200 pixel resolution vision sensor captures full-frame, two-megapixel images at up to 14 frames per second, said to provide image stability for more reliable analysis when viewing moving objects. It provides high accuracy for precision gauging, alignment, and inspection of smaller parts. Increased resolution of the 5403 can also be used to provide a larger field of view for capturing images of big parts, or viewing multiple objects simultaneously. Alternatively, the 5403 can be mounted further from the part and still achieve required resolution. It is a self-contained, industrial-grade vision sensor that meets IEC specifications for shock and vibration, and achieves an IP67 (NEMA 6) rating for dust and wash-down protection. It offers built-in Ethernet communications, and In-Sight Explorer software combines the unique and powerful vision spreadsheet interface with advanced tools to simplify application development and network administration. Running with optional PatMax object and feature location tool, 5403 is expected to significantly improve accuracy and repeatability of vision-guided motion, and similarly demanding systems. www.cognex.com Cognex

Vision sensor has 1.3 megapixels & $2k

PresencePlus P4 Geo 1.3 vision sensor is said to enable highly detailed automated inspections at about half the cost of comparable vision sensors. It uses a 1.3-megapixel imager to capture minute details of multiple features at ranges from a few inches to several feet, depending on lens used, so it can perform an inspection that previously required multiple sensing devices. In automotive, packaging, pharmaceutical, assembly, metal forming, semiconductor manufacturing and plastics, it can verify that a robot correctly dispensed a bead of adhesive around the edge of a car door’s interior panel; check that multiple widely spaced mounting clips are in the correct locations and aligned properly; and ensure a box has 24 bottles before case is sealed. Remote teach feature allows the sensor to learn new features to inspect, without connecting to a PC or shutting down the production line. A built-in live video image output allows users to view the sensor’s inspections in real time and to view failed inspection images without a PC. It has a built-in 10/100 Ethernet connection, RS-232, or four discrete input/outputs. Three bi-color LED indicators show unit status during setup and operation. It operates on 24 V dc; hundreds of lighting and C-mount lens options are offered. www.bannerengineering.com Banner Engineering

2 megapixel CCD vision system

CV-2600 high-speed machine vision system has a 2 megapixel (1,620 x 1,220) CCD, and incorporates numerous ‘first-in-class’ features, one of which is the 2 mega-pixel CCD built into the industry’s reportedly smallest camera housing. The 2 Mega-Pixel CCD [charge-coupled device, which is chip-based memory activated by light] can inspect an area 4 times larger than conventional cameras at the same resolution, and with a low-voltage differential signal transfer method, operates 2.5 times faster than conventional models. This saves significant installation time and cost. CV-2600 controller supports 2 cameras. Its super small, double-speed, progressive-reading camera has wide mounting/application capabilities. Camera sensitivity, adjustable in 81 increments, eliminates strobe lighting, and allows high shutter speeds. High-Flex Camera Cable, capable of over 1 million bend cycles (10x conventional cables), suits robotic use. Inspection tools include area, pattern search, multiple search, edge angle, edge width, number of edges/pitch, stain, blob, intensity, trend edge position, and trend edge width. On-screen programming menus guide users through simple setup; no PC is required; and configuration is said to be easy. Ultra-high-speed, real-time rotational searches can be done at minimum speed of 61 ms (about 1,000 parts/min.). Repeatability is www.keyence.com Keyence

Advanced algorithms speed machine vision

NI Vision 7.1 Development Module features hundreds of vision tools for engineers and scientists using NI LabView, LabWindows/CVI, C/C++, or Visual Basic to create powerful vision applications that inspect, align, identify, and measure objects. Software allows access to new algorithms, such as geometric matching, object classification, optical character recognition (OCR), and 2-D bar code readers, to quickly locate, sort, and track objects even in poor imaging conditions. Geometric matching locates objects by edges and shapes, rather than textures and shades, and helps when objects appear to change scale, as with robotic guidance and product assembly, where the distance between the camera and object varies. Other tools help with part identification, tracking, and sorting. NI Vision Assistant allows application prototyping before code writing, then converts script to LabView diagrams or ready-to-run NI LabWindows/CVI, C/C++, or Visual Basic code. Engineers can run the code or add it into a larger machine vision, industrial control, or data acquisition application.

Separately, NI Vision Builder for Automated Inspection 2.5 is configurable software for building, benchmarking and deploying machine vision applications without programming, and with new algorithms, industrial connectivity, and improved multi-camera support for its machine vision applications. Vision Builder is priced from $1,495. www.ni.com/vision National Instruments

Smart vision sensors lower cost, increase speed

A compact 2-D smart vision sensor, the ZFV, is designed to bring higher sensing speeds, greater efficiency and lower operating costs to sensing and vision applications. Speeds of 4, 6 or 12 m/s can be selected, and intelligent lighting optimizes image quality and ability to adjust focus maximizes measurement accuracy. Images are transmitted digitally from sensor head to an amplifier and between amplifiers for clearer images. Sensing view is 5 to 50 mm. Single-function or standard multifunction amplifiers are available to fit the application. Advanced algorithms expand viewing areas as well as enhanced discrimination of pattern, brightness, character detection, label position, target width, and detecting bar codes. Operations are selected on the amplifier’s LCD monitor using easily identifiable icons. The LCD displays real-time inspection images and can be panel mounted to eliminate separate monitoring equipment. One-touch tracking simplifies set-up. ZFD is said to be ideal for packaging and electronics operations where cost-effective high-speed inspections are necessary. Price is $2,000 for a ZFD sensor/amplifier system. www.omron.com/oei Omron Electronics

Flexible, easy vision system follows standard

NetSight II DCL is a Camera Link version of the NetSight II machine vision system, which processes multiple image views simultaneously. Designed for applications that require line or area scan digital cameras, NetSight II DCL allows users to match the image of choice to their inspection needs. Camera Link is an interface standard said to use a dedicated cable connection and is designed for demanding environments. Users get unprecedented flexibility in a variety of applications and ease of installation and integration, in a rugged, cost-effective package, the company says. System provides asynchronous acquisition of images from up to two Camera Link cameras (area or line scan formats) and one FireWire camera using an IEEE 1394 port. It has a high-performance processor, 256 MB of program memory, 20 GB of storage memory, and programmable digital I/O connections. Built-in Ethernet connectivity and compatibility with industry protocols like EtherNet/IP and Modbus makes it easy to integrate. NetSight II DCL uses the Sherlock vision software as its interface, said to be a powerful inspection tool, with easy graphical interface for fast prototyping and deployment, configurable for any application; it includes inline scripting. www.goipd.com ipd div. of Coreco Imaging

Intelligent cameras make a system; free tutorial

Impact T25 and T21 are the newest additions in the T-Series intelligent camera family for high-speed automated inspection. T25 is designed to give manufacturers greater accuracy and improved defect detection compared to standard-resolution vision systems. T21 is a remote micro-head-based camera, for tight spaces. Both are full-featured, self-contained vision systems. Online are a full-function software demo, and free online tutorial details impact on manufacturing quality and throughput. Inspection Builder software contains all features of a full-scale vision system, deployed in an economical smart camera format. PPT says its fully digital systems capture pristine images that can be transmitted, noise-free, faster and over great distances, eliminating many limitations of conventional analog systems. www.pptvision.com PPT Vision

Remote inspection

DualVision 724 Remote Monitoring System for electrical substations is said to be first to combine visual and infrared (IR) cameras to produce optimized, blended images for simple quick analysis within one intranet/Internet-enabled package. It finds temperature excursions on substation hardware and site intruders, and can trigger alarms. MikroSpec R/T software is used to produce a composite IR and visual image, and separate images of each. Composite can be viewed in an infinitely blended percentage of visual/IR, by moving a slider bar in the software. Up to 32 regions of interest can be defined on the thermal image in any complex shape. www.irimaging.com Mikron Infrared Inc.

Online Extra Link to the repository of Control Engineering product research reports at the Resource Center, a specialized micro-site dedicated to engineering professionals. (Subscription required.) Click here to view the repository .https://resource.controleng.com/community/23084/Monthly+Product+Focus.html

Another poll : In a separate poll, during February, Control Engineering Online asked readers in a one-question poll at www.controleng.com, “When was the last time you evaluated machine vision for an application?” See www.controleng.com/poll for these and other results. Because these are open to any visitor, respondents’ buy/specify influence cannot be confirmed, unlike subscribers that respond to Control Engineering “Product Research” surveys.

Online Products 2,100-ft infrared vision Thermal-Eye TSC Series cameras use heat, not light, to sense temperature deviations from as far as 2,100 ft. Applications include industrial process monitoring, preventive maintenance, and security. A medium-range lens expands field of view up to 1,000 ft. A wide-angle TSC camera offers detection up to 300 ft. Infrared cameras need no lighting. Infrared Products (formerly Raytheon Infrared)

Vision measures CNC parts quickly Ultra Quick Vision Series CNC non-contacting measuring systems are said to employ advanced construction techniques to achieve extreme accuracy (U1XY = [0.3+L/1000]xpansion as a factor affecting accuracy. A four-quadrant, programmable ring light allows easy adjust of illumination. Mitutoyo America Corp.

Capture up to 200 frames/s As a dual-tap AccuPixel offering, TM-6740CL boasts up to 200 frames per second at full resolution and generates up to 3,205 frames per second in partial scan and binning modes. This 1/3-in. VGA-format progressive scan camera is based on the state-of-the-art Kodak KAI-0340 CCD imager. The camera’s 640(H) x 480(V) pixel resolution imager has 7.44,000 s or pulse-width exposure control, which allows triggered image capture and processing. Read-out-inhibit control makes multiple-camera applications possible. Housing is 50.8 x 50.8 x 85.1 mm. Other features include a software GUI, automatic dual-channel compensation, and a built-in pattern generator. Jai Pulnix Inc.

Fast, wide vision acquisition Odyssey Xpro+ scalable vision processor board features the latest version of Freescale G4 PowerPC embedded microprocessor with a core frequency starting at 1.4 GHz, a state-of-the-art Oasis processing and router ASIC, and a customizable co-processor FPGA (field-programmable gate array). Single-slot Odyssey Xpro+ also offers over 5 GB per second of memory bandwidth, up to 2 GB of DDR SDRAM memory and up to 2 GB per second of external I/O bandwidth. These features, combined with PCI-X bus technology and linearly scalable architecture, equip Odyssey Xpro+ with the power and flexibility needed for the most demanding vision applications. Matrox Imaging

Vision guides robotics, increases quality TrueView Vision Guided Robotic Systems combine ABB’s robotic expertise and Braintech’s software platform, eVisionFactory (eVF). TrueView allows manufacturers to configure assembly lines in a truly flexible and adaptive way. eVFTM integrates hardware components (robots, robot controllers, end effectors, and calibration devices) with guidance system software, allowing manufacturers to increase ability to adapt an automated assembly line to changing materials handling requirements and other tasks that robots are designated to perform. Braintech

LEDs enlighten machine vision X-Beam is a rugged, high-intensity, LED beam-lighting system, available in a variety of colors. Optics resolve most of the beam consistency issues common to LED lighting. It can be used with either line or load-side switching, depending on application speed. X-beam connects to power supply via a circular, 4-pin, water-resistant connector. Size is 1.5 w x 1.75 d x 1.5 h in., in configurations with up to five LEDs. It has an aluminum housing, with several mounting options. FSI Machine Vision


Author Bio: Mark Hoske has been Control Engineering editor/content manager since 1994 and in a leadership role since 1999, covering all major areas: control systems, networking and information systems, control equipment and energy, and system integration, everything that comprises or facilitates the control loop. He has been writing about technology since 1987, writing professionally since 1982, and has a Bachelor of Science in Journalism degree from UW-Madison.

Related Resources