We review and interpolate numeric information all the time to create a virtual image of the condition so we can relate to the information provided. Thermal imaging is a little different in the role of the imager because the way in which information is captured through the creation of the image, we are supposed to believe it. However, can you? Do you trust what you see? How accurate is the information presented?
There is a triad of technologies we use every day: ultrasound, vibration and thermal imaging. Using all three together is one of the best ways to get all the facts and make the best determination. The key is understanding the technology behind each instrument. Let’s focus on the use of thermal imagers. With infrared (IR) technology, we are actually measuring the infrared radiation coming from an object’s surface and converting this data into a temperature measurement. In a fraction of a second, it is assembled by the imager using a complex set of algorithms into a visible image for us to review, create and base an informed decision on. For this article, let us assume ideal conditions are present for thermal imaging, such as no external IR source interference and correct emissivity settings. We will then use this visual image in one of three ways, which are the most common, to create this informed decision. These three typical scenarios are:
Scenario #1 - Temperature difference between the object and ambient conditions.
Scenario #2 - Temperature difference between the object and a similar object under the same load conditions.
Scenario #3 - Actual temperature of the object and known safe operating temperature range of the object.
The first two are the most common of these three, as over time, you will build up experience and comparison data based on many inspection points and operating variables. Nevertheless, how can you trust the temperature provided? Can you believe what you are seeing?
The way to ensure accuracy is to understand thermal imaging principles. I highly recommend a basic thermal imaging Level 1 course from one of the many qualified providers. In these courses, they will instruct you on the variables and how to limit the potential impacts and effects they may have on your data and the operation of your thermal imager. The second part is having the right equipment at the right time for the job at hand.
Resolution and sensitivity are two very important specifications, followed very closely by lens selection and software capabilities. All these characteristics working in unison are the components that make up your imaging platform. Having one or two specifications and lacking others will make your task of complete and comprehensive inspections very difficult and cause more headaches down the road. In many applications, to get accurate and reliable temperature measurements of your area of interest, resolution is the key factor.
Better and higher image resolution creates a positive condition that is directly proportional to providing more data collected to build the thermal image, provide more detailed and clearer representation of the measuring object and give accurate temperature measurements. There are two typical ways to get more resolution. One is to buy a larger imager with higher resolution. This may or may not be a viable option depending on your work environment and operational conditions. The second method is to figure out a way to completely fill the area of the sensor with the most data possible by getting as close as possible to your area of interest to maximize the coverage of your imager. This may not always be possible either, because of location or safety requirements in place within your plant or facility operation.
Most equipment manufactures recognize these issues and work towards developing technologies that address this issue directly by improving the usable, geometric resolution of the thermal image. Recently, an image enhancement technology has been introduced in which thermal resolution increased up to a factor of 1.6 times (1.6X), or 60 percent, which increases the resolution (pixels) by a factor of four times (4X). Let us review this process in more detail as the idea behind the process is simple, yet the actual development of this technology has been quite complex.
The inherent issue is the layout of the pixels in a thermal array. To prevent thermal energy from being conveyed from one to another, individual pixels cannot be touching. This would provide very poor and inaccurate measurement and an imaging system that is not very effective. Since the individual pixels of any sensor have these areas of dead space in between them to act as thermal insulation between the pixels, these blank areas do not record any measurement data. However, what if there was a way to fill in these blank spots with real temperature measurement data, not a simple interpolation based on scaling up of the image? This filling in of real data is exactly what happens when the imaging process uses a multi-image capture protocol in generic terms of multi-image sampling.
Starting with a rapid collection of images, five in total captured within 200 ms, then using a complex image registration process where the first image serves as the base image. then images two through five are compared pixel by pixel. When the data is confirmed to fall within the dead space of the main image pixel matrix of image one, then and only then is the data inserted from the other four images captured into the matrix to completely fill it in with this real data. This effectively increases the real pixel count by the factor of 4 times (4X).
Table 1 - Pixel Count Increase of Process
Original Pixel Matrix
160 x 120
320 x 240
320 x 240
640 x 480
640 x 480
1280 x 960
The next step in the process is called deconvolution, the basic function to optimize the image through an in-depth understanding of the optical properties of the lens and imaging path being used. The end result is an effective increase in the instantaneous field of view (IFOV) of the imaging system by having more pixels within the image than could be captured by the original pixel matrix. This provides a superior data set that provides additional pixels on any given area within the image that was not possible due to the size of the original matrix. This additional data provides the ability to create a much improved thermal image with exceptional temperature accuracy as a direct result of this increased level of data availability. The overall temperature accuracy of the thermal imaging system can be improved up to a factor of 1.6 times (1.6x) or 60 percent over similar imagers not using this technology.
Table 2 - Imaging System IFOV Increase of Process
IFOV ~ 1 meter
IFOV ~ 1 meter
160 x 120
320 x 240
320 x 240
640 x 480
640 x 480
1280 x 960
The actual increase in resolution and IFOV can be tested using a modulation transfer function (MTF) of the imaging system. This test shows how image resolution is increased, providing not only sharper, better image clarity, but also improved temperature accuracy by extending the measurement area before fall-off occurs.
How can this be seen or can it be seen in real life images from actual inspections?
In the following example, captured during an inspection of a substation, an overhead knife switch was detected to be operating in a hot condition when compared to other switches under similar loads nearby. This goes back to scenario #2 of how we use thermal imagers. The initial temperature of the switch is 109°F, but when enhancement technology is applied to the imaging solution, a new, more accurate temperature of 113°F is recorded, as well as a highly detailed sharper image. This 4°F temperature delta did not represent a critical repair, but it did elevate the switch to a higher level of inspection to monitor the temperature more frequently.
Overhead Knife Switch - initial thermal reading and new reading once enhanced technology is applied.
New and exciting technologies are under development every year by equipment manufacturers to increase the capabilities of your testing equipment. These technological advances provide the ability to capture images and trust in the data collected to build better and more comprehensive CBM and reliability maintenance programs for your operations. You should place trust in your measurement equipment, but only after you take the time to understand how and what is measured. Keep in mind the most important piece of the equation is you and your interpretation of the information presented.
Kevin R. Lesnewski, North American Product Manager, Thermal Imaging Division, Testo Inc. With over 20 years of experience in imaging science and imaging applications, and formerly a Senior Imaging Specialist with the Kodak Scientific Imaging group, Kevin has extensive and specialized experience in many non-conventional imaging applications and technologies used in the fields of research, manufacturing and process control applications. www.testoUSA.com