What industrial camera would be best for taking pictures at a chemical reactor?

Charlie Masi

What industrial camera would be best for taking pictures at a chemical reactor?
November 4, 2008

I can’t be too specific about a recommendation for reasons that will become clear later. There are is a wide selection of cameras, each adapted to a fairly narrow range of applications. It all depends on what you want to do with it.

There is no “general purpose industrial camera” that is appropriate for all industrial applications. Off hand, I can think of about five classes of digital cameras of interest in industrial applications. I’ll give you a rundown on all of them, so you can choose what most closely matches what you want to do:

Industrial photography — Handheld or tripod mounted cameras equivalent to the 35-mm SLR cameras so loved by professional photographers everywhere are still around. They just capture images on CCD or CMOS based image sensors instead of photographic film. Yes, there are still film cameras manufactured, but they are specialty items. Kinda like buggy whips. Their control systems for the three parameters photographers need to control (focus, exposure time, aperture) reached a high level of sophistication before image sensors replaced film, and they have continued developing since. Look for a standard lens-mount system that will allow you to couple the camera body with a selection of third-party lenses. When I bite the bullet and replace my beloved Canon AE-1 film camera, I’ll almost surely pick a model from the company’s Rebel series. If price were no object, however, I’d look into Nikon’s D series. My experience (and apparently that of most professional and semi-professional photographers) has been that Nikon professional level cameras somehow make you take better pictures. Use these cameras to take pictures of what you’re doing, but do not, generally, use them as part of your process.

Surveillance — My introduction to video cameras came in college when I had a summer job as an electronics technician. One project was installing a set of video cameras to observe activity on a large production floor. I’ve no idea what they planned to look for, but the employees gave us a lot of dark looks as we carried a bunch of closed-circuit TV (CCTV) cameras up into the rafters to be aimed at them. In general, surveillance is still done by CCTV cameras wired directly through recorders to monitors that underpaid humans stare at. The cameras are, for the most part, junk compared to anything in the machine-vision world. More sophisticated technology is available, and may slowly be adopted, but cheaper-is-good-enough seems to be the watchword.

Automated inspection — Probably the most common use for machine vision in industry is still automated inspection for product quality control. Because this market is so big, there are a large number of products offered with characteristics optimized for specific niches, such as circuit-board inspection, mechanical component dimensional inspection, print inspection, color, etc. I hesitate to even suggest who the technology leaders are. Things change faster than I can keep up. Such cameras are most often black and white using fixed optics. They are generally permanently mounted in fixtures that allow exquisite control of field of view, product placement, lighting, etc.

Motion-control and robot guidance — As automated production systems become more common, the need for flexible machine guidance sensors becomes more acute. Cleverly designed machine-vision systems using sophisticated high-speed image analysis software coupled with motion-control systems are becoming more popular. In this category, the camera often takes over for a number of more traditional sensors. For example, machine vision can measure positions and speeds very accurately. I’ve seen applications where machine vision helps two robots fit components together as a human assembler coordinates two hands. Certainly all vision-based driver aids making their way onto highways fall into this category. The application list is long and growing.

High-performance imaging — The most expensive cameras with the most stupifying performance specifications serve specialized applications. The best way to communicate what I’m driving at is to describe a few applications I have in mind.
* A number of sports-equipment manufacturers use high-speed, high-resolution cameras to study what happens when a stick hits a ball. Baseball, hockey, tennis and other stick-and-ball sports have been extensively studied this way.
* Web-inspection systems can locate flaws in, for example window film, flying past at hundreds or even thousands of feet per minute. At the other end of the spectrum, cameras literally watch paint dry for product-development purposes.
* Optical astronomical telescopes are basically specialized cameras that can pile up photons one at a time to form an image over tens of hours of observation time, or take quick snapshots of thousands of objects per night. Most such cameras are fully automated pieces of equipment designed for a specific project at sometimes enormous expense. Even amateur astronomers have systems available that can take a typed-in name of an object to observe, look it up in a star catalog, find it, autofocus on it, autoguide to avoid motion blurring, and shoot the picture sans human intervention.

There are a few companies that provide mostly off-the-shelf units that fit a range of applications. In any case, for high-performance applications you should begin consulting with the camera manufacturer as early as possible, giving people there as much detail about your application as possible, and let them help you work out the specifications.

This is almost surely an incomplete list, and the categories are pretty broad. Nonetheless, I think it hits 80-90% of the market. I’ll bet your application fits in one of them.

Finding the right camera

From the question’s wording, I suspect that the application in mind fits into the last — high-performance imaging — category. It sounds like the questioner wants to observe something visual about the process. Maybe it’s a color change that occurs as a reaction proceeds. Maybe it’s a question of how two reagents mix prior to reaction. Perhaps it’s an exothermic reaction that proceeds at high speed, and the engineer wants to watch a flame front move through the reaction vessel.

In any case, here are a few steps to take to get the best result:

1. Clearly define what you’re trying to observe, along with any constraints. Especially think about time, available light, and the space you have to work in. Think very carefully about how what you want to see provides contrast for the vision system. Remember, redheads look like brunettes on a black-and-white TV. Cameras can sometimes see things that human eyes can’t, and they can completely miss things we see clearly. These considerations all fall under the heading of “what you want to do.”

2. Study available tutorial sources. Edmund Optics has an especially useful online library for beginning through advanced machine vision engineers. For information specific to automated control topics, visit the Control Engineering Website and search on machine vision as a keyword. While you’re there, visit the Resource Center for tutorials and white papers. Use these sources to help you define “how you want to do it.”

3. Seek out advice from machine vision vendors and system integrators. Check the Control Engineering System Integrators Guide for individuals and companies with experience solving problems like yours, and visit the Buyer’s Guide to find machine vision equipment suppliers. Use these sources to help you plan “how to make it happen.”