Infrared-Based Cameras and Machine Vision

The early days of infrared imaging were driven by military objectives – night vision and the like. It is probably still true that the latest advances we have been seeing in IR cameras have also been driven by the military. Fortunately, machine vision can now take advantage of these advances, which have yielded cameras more consistent with machine vision applications.

Features of these new cameras include higher resolution, higher frame rates and room temperature operation (i.e. no need for coolers) as well as lower cost. Significantly these cameras will have spectral bands of operation – near infrared, mid range infrared and far infrared. Certain spectral bands are more suitable for certain applications.

It is still not apparent that the industrial applications of these IR cameras are really machine vision where by ‘‘machine vision’‘ we are implying unattended operation, or operation without sustaining human intervention. Many of the applications cited relate to machine diagnostics and generally there is a person viewing a display of the IR image or thermal image making a judgment call based on what he/she sees. Some of these cameras do come with software designed to automatically characterize the IR or thermal data. Some of these cameras are digital and consequently their output can be fed directly to a personal computer, where imaging and machine vision software can be used to operate on the image.

A number of applications have been cited including mold or form inspection, printed circuit board inspection, glass – hot gob – inspection, weld seam inspection, package seal inspection, thermal grid inspection (wiring integrity in automotive rear window glass) and silicon wafer inspection.

To gain further insight into the latest developments in infrared and thermal imaging cameras, suppliers of such cameras were canvassed for input to this article. The following responded and provided answers to our questions:

• Brooke Herbst – Marketing Communications Coordinator – Electrophysics
• Rich Barton – Product Manager – Optotherm
• Arn Adams – Product Manager – Santa Barbara Focalplane/Lockheed Martin

1. How would you describe your product line that specifically addresses IR/thermal machine vision applications in manufacturing? 
[Brooke Herbst – Electrophysics] Electrophysics PV-320 is a high-resolution thermal camera with accurate temperature measurement and a USB 2.0 real time digital output.

[Rich Barton – OptoTherm] OptoTherm’s EL product line enables printed circuit board manufacturers to detect and locate defects such as short circuits and stressed components.  The R&D product line allows manufacturers to create custom tests to evaluate the thermal behavior of their products.

[Arn Adams – Lockeed Martin] We sell high performance cryogenically cooled thermal IR camera systems tethered to computers.

2. What specifically differentiates the products you offer? 
[Rich] By focusing our efforts on electronics defect analysis, we can offer customers in this industry solutions that address their specific needs.

[Arn] Our thermal IR camera systems allow streaming data to disk at high data rates, are the highest performance of any systems (highest data rates, fewest dead pixels, most uniform, cover the broadest wavelength range, and have highest sensitivity) and are supported by an easy-to-use Software Development Kit. 

[Brooke] We offer products that cover the entire infrared spectrum and offer a wide range of price and performance levels.

3.  Would you characterize IR/thermal-based machine vision systems as ‘‘hardware-centric’‘ or ‘‘software-centric’‘ and why? 
[Arn] SBF's thermal IR cameras fit well in ‘‘software-centric’‘ systems because of our ease of integrating with nearly any computer system. 

[Brooke] Our solutions are hardware-centric because we offer software developer kits (SDKs) to support third party developers. We believe integrators know the applications much better and they also take total systems responsibility.

[Rich] Both. Hardware centric - an IR camera with sensitivity in a specific wavelength band may be required for testing certain materials.  Software centric - many tests require custom algorithms to appropriately evaluate a product or material.

4. What are some machine vision applications related specifically to manufacturing that you have addressed with your products? 
[Brooke] There are many. We have supplied auto assembly plants thermal cameras to test defrosters in windshields for shorts in electrically conductive coatings. We supply cameras to a company that shreds cars and other large metal objects. They use steam to keep the potential for explosions to a minimum and thermal can see through the steam and allows them to determine the correct operation of the shredding mechanism because a visible camera cannot see the process.

[Rich] Identifying defects on power supplies, detecting cracks in ceramics, evaluating the thermal distribution on glass plate.

[Arn] Our thermal IR cameras have been integrated into a wide range of machine vision systems by other vendors, especially non-destructive test applications and FTIR spectroscopic examination of materials. 

5. What are the critical parameters of an IR/thermal-based machine vision application that a prospective customer should understand? 
[Rich] Make sure that the wavelength of operation is appropriate for the application.  Long IR (7-14 microns) is appropriate for the vast majority of manufacturing applications.  However, some applications may require the use of short IR (3-5) or mid IR (5-7 microns) if the materials being tested are transparent or reflective in the long IR region.  Also, make sure the software has the ability to create the kind of test that is required.  For example, the software may need to analyze the difference between the mean and minimum temperature within an area in order to identify a material crack.  Additionally, make sure the camera can perform the tests at the required production rate and the software can supply the information that is required by other computers, machines, or a network.

[Arn] Each customer has different requirements, but generally they want the best image, easy-to-use software development kit, and high data rates streamed to disk or RAM.

[Brooke] If they are interested in measuring temperature they should understand the ambient temperature compensation of the system under evaluation. How will they control and interface to the camera? If the application requires inspecting objects in motion, what is the effective exposure time of the camera?

6. What are the skills required to program and operate an IR/thermal-based machine vision system? 
[Arn] The skills required are highly dependent on the particular application. SBF doesn't sell specific machine vision systems but our customers often comment on how easy it is to use our highly optimized software development kit for their machine vision applications.

[Brooke] They are the same as visible cameras.

[Rich] Programming requires familiarity with operating Windows based software applications.  Extensive HTML help shows users how to setup and use all of the software features and capabilities.

7. How do you support your products – training, documentation, warranty, post-installation service, software revisions? Are these free or is there a fee?
[Brooke] We provide detailed ICD documentation. Software revisions are free for within-release fixes for bugs. Since we primarily sell drivers and they are stable, we haven’t seen a great deal of demand for support and upgrades.

[Rich] Each system is shipped with a full set of printed hardware and software manuals and includes a 1-year standard warranty.  We also offer onsite training and installation for an additional fee.  Each system also includes one-year free software upgrades.

[Arn] Most customers rely on our documentation for their training; on-site training can be provided for an extra fee, or free training is available at SBF.  Software revisions are available for a fee after the first year.  Typical warranties are 6 months to 1 year, depending on the product.

8. Where do you see breakthroughs coming in the specific infrastructure technologies (hardware and software) that are the basis of IR/thermal-based machine vision systems that will result in further improvements in the near future – next three years? 
[Rich] The continued improvement in detector sensitivity and optics will allow much smaller devices to be analyzed.  The development of application-specific testing methodologies and software algorithms is required to provide solutions for many new IR manufacturing applications.

[Arn] The introduction of our digital output Focal Plane Arrays has provided an important breakthrough in thermal IR technology.  Also, new detector materials, for the first time in several years, look to be very promising in the near future.

[Brooke] The most important development is lowering price points. We already have today industry standard interfaces like USB, Camera Link and FireWire.

9. What specific performance improvements are anticipated driven by these forthcoming technological changes? How will they impact the use of these systems? 
[Arn] Digital output Focal Plane Arrays will improve detector sensitivity and uniformity, simplify the electrical interface to the FPA, reduce power requirements, make the camera more immune to external noise sources, and improve by more than a factor of 10 the characteristics of integration time turn-on and turn-off (important for high speed events).

New detector materials offer the possibility of higher performance systems that work at much higher temperatures than the cryogenic sensor temperatures that are currently required.

[Brooke] Performance will increase in terms of resolution but costs will increase for those systems.

[Rich] The hardware and software technology is already at a level at which this technology can be integrated into many manufacturing applications.  Lack of a general understanding of IR fundamentals has prevented the widespread incorporation of this technology in manufacturing.

10. What are the main markets for these systems and are there market changes within those markets that are driving the adoption of IR/thermal-based machine vision systems? 
[Brooke] The market for infrared is still developing so I do not feel there is a main market yet established. As the price points fall, the technology will begin to become more attractive and the adoption of IR more universal.

[Rich] Main markets include electronics manufacturing, soldering, material NDT, insulation testing (car door, for example).  The primary market change that is motivating manufacturers to adopt this technology is the price reduction of IR systems.  The reduction in camera and software prices, however, has leveled off in the last few years.  The adoption of IR for manufacturing will continue to be slow and steady, because an understanding of IR fundamentals is required for manufacturers to incorporate this technology into their process. 

For this reason and because a history of IR information has yet to be developed for the vast majority of manufacturing applications, the use of IR in manufacturing will continue to be limited to the most straight forward manufacturing projects in the near future.

[Arn] I'm not able to comment on new markets or how markets are changing without giving away proprietary information.

11. What impact do these market changes have on the requirements for IR/thermal-based machine vision systems? And how will machine vision systems have to change to address these more demanding requirements? 
[Rich] IR system manufacturers and providers must provide whole-product solutions for their customers and must be willing to take the time and effort to educate their customers as to the uses and benefits of IR and how to properly incorporate IR systems into their process.

[Arn] The market for new machine vision systems will require higher performance sensors, often at lower cost than current systems.

12. As a supplier of thermal/IR-based machine vision-based systems what are some of the challenges you face in marketing machine vision systems? 
[Arn] SBF doesn't market or sell machine vision systems, only thermal IR camera systems.

[Brooke] We can only offer a sensor and not a turnkey solution. There are a limited number of integrators with any experience in thermal imaging.

[Rich] The largest challenge is making sure our customers are adequately educated and trained to take advantage of the benefits that IR can provide.

13. What advice would you give to a company investigating the purchase of an IR/thermal-based machine vision system? 
[Brooke] If possible conduct a live on-line demonstration of the camera. If the logistics are too difficult consider renting a system for a period of time and invest in the housing and installation of the rented system. Most manufactures offer credit of rental payments on future purchases.

[Rich] I would advise purchasers to request an onsite demonstration of the system.  The demonstration should include testing the specific products/materials for which the system will be used.  For the majority of applications, the system software and hardware can be setup by the sales representative/application engineer during the demonstration to perform an analysis of the products/materials similar to what is required in the real-time manufacturing environment.

[Arn] I always tell potential customers to seek out the best system for their needs, which may not be a thermal IR camera system by SBF.  I try to stay abreast of changes in the industry to help steer customers toward the best solution for their needs.