Different Factors Which Impact Thermal Imaging Quality

Thermal imaging capability can really make a product design stand out from the crowd. As well as offering the ability to see clearly in complete darkness and without any additional illumination, thermal cameras are able to detect and display the subtle temperature differences of all the objects within the camera’s field of view.

Temperature differences collected via thermal imaging can be used to highlight qualitative differences in relative temperature. For example, this can show that one set of objects is warmer or cooler than other objects in the vicinity; or it can allow users to gather accurate, non-contact measurements of temperature. Not only that, but a thermal camera can actually do both of these things at once, allowing the tracking of temperature trends over time.

Not every application actually needs quantitative temperature measurements, and qualitative imagery can show, for example, the coldest or hottest objects on the scene, colorizing these to alert the user to temperatures that may exceed a specific level or threshold. Security cameras and firefighting applications use this functionality within their day to day operations.

Products can be created to gather accurate measurements of temperature, but a number of factors much be considered as these can seriously affect the accuracy of any measurements taken. The following sections provide an overview of the most common of these factors.

Radiometry and Surface Characteristics

Radiometric thermal cameras measure the temperature of a surface by interpreting the density of an infrared signal emanating from that surface and reaching the camera itself. This is a non-contact and Non-Destructive Technique (NDT) that offers tremendous advantages over many others; in particular the ability to measure surface temperatures from a distance, helping to not only ensure the safety of the person operating the camera but also making sure that the item being measured cannot be influenced by the measuring instrument itself.

Camera operators need to address unique radiometric temperature challenges when gathering accurate temperature measurements.

Figure 1. Camera operators need to address unique radiometric temperature challenges when gathering accurate temperature measurements.

It is important to note, however, that in order to sense temperature remotely (as opposed to via direct surface contact), the user must be able to account for a variety of surface and environmental conditions.

Radiometric temperature measurements are generally made of an object’s surface specifically because they are made up of optically opaque materials – organic materials and metals, for example, are usually totally opaque within the infrared spectrum.

Some materials are, however, semi-translucent to infrared heat such as zinc sulfide glass, sapphire glass or zirconia oxides. Radiometric measurements of these surfaces should account for the total volumetric and through-plane temperature of the material. Because accounting for this makes the analysis much more complex, the majority of measurements are made of opaque materials.

In order to sense the temperature of an object’s surface remotely, one must rely on the ability to accurately compensate for characteristics of the object’s surface as well as those of the imaging system itself.

Specific surface characteristics are able to influence the temperature measurement; emissivity and reflectivity at the infrared spectral wavelengths. Additionally, the atmosphere will both absorb and emit thermal energy based on its ambient temperature, composition, humidity and the distance between the camera and the surface.

Lastly, the ability to actually spatially resolve detailed temperature measurements within a thermal image is influenced by three key things: blur, image focus and pixel resolution. Exactly how each of these factors impacts on the accuracy of measurements depends on the specific measurement application, but these must be accounted for and resolved in order to maintain image quality.

Emissivity

Emissivity is a measure of how efficient a surface is in emitting thermal energy. This is measured relative to a perfect emitter known as a ‘blackbody’. This measurement directly scales the intensity of the thermal emission with all real values being less than 1.0.

In most cases, the emissivity of a particular object is dictated by the materials that it is composed of, as well as its surface texture. Organic materials are generally highly emissive, for example, soil, human skin, unpainted or varnished wood or rocks.

The emissivity of objects is very much dependent on their surface compositions, but this can also be affected by things like the object’s oxidation, roughness, temperature, spectral wavelength or even its viewing angle. Polished or shiny materials are more reflective, on the other hand.

It is important to be aware of all these factors and account for them appropriately, as any measurements that do not do so run the risk of surfaces appearing to be colder than they actually are.

In agricultural applications, a lot of organic materials and materials that have rough surfaces are considered to have emissivity values approaching 1.0. Other applications such as solar cell or power line inspection will see surfaces that are made of polished glass or metal, and these will have far lower emissivity values.

Table 1. Emissivity values for common materials. Tabulated for reference only; emissivity is dependent on surface morphology, layers, oxidation, spectral wavelength, view angle, temperature and view angle.

Material Description Emissivity [n.d.] 1, 2
Asphalt 0.90 to 0.98
Concrete 0.92
Soil, dry 0.90
Soil, wet 0.95
Wood 0.90
Water 0.92 to 0.96
Ice 0.96 to 0.98
Snow 0.83
Brick 0.93 to 0.96
Lacquer, paint 0.80 to 0.95
Lacquer, flat black 0.97
Textiles 0.90
Skin, human 0.98
Aluminum, polished 0.04 to 0.06
Aluminum, anodised 0.55
Steel, rusty 0.69
Steel, stainless 0.16 to 0.45

Every thermal image includes materials with different levels of emissivity and reflectivity. Understanding these properties is vital for accurate temperature measurements.

Figure 2. Every thermal image includes materials with different levels of emissivity and reflectivity. Understanding these properties is vital for accurate temperature measurements.

Reflectivity

When a camera is close to a surface, it is often sensing heat from the surface temperature itself, but also reflected temperature from the background environment. Measuring the temperature of very reflective surfaces is particularly challenging because of these background thermal reflections; for example, when taking measurements outside, a clean, unpainted metal roof may appear colder than it really is because it is reflecting the sky above it.

In order to give this concept a practical, demonstrable example, it is possible to consider the case of a stainless steel sheet on a rooftop. The sheet has a reflectivity of 0.80 and an emissivity of 0.20.

When using radiometric temperature measurements in these conditions, the result would be highly biased to the background temperature of the sky as it is reflected from the stainless steel sheet.

A clear sky could have a background temperature that is well below 0 °C, even as low as -20 °C. The background temperature of the sky can vary greatly depending on the time of day or the current atmospheric conditions.

Radiometric temperature measurements should avoid straight-on measurements when imaging highly reflective materials to reduce direct camera reflection and avoid oblique angles to reduce overall reflection.

Figure 3. Radiometric temperature measurements should avoid straight-on measurements when imaging highly reflective materials to reduce direct camera reflection and avoid oblique angles to reduce overall reflection.

In a similar fashion, the sun’s reflection in the thermal image will appear as sun glints; the radiometric temperature of these sun glints reading as hundreds of degrees different to the actual temperature. In order to manage this, it is generally recommended to take a series of images of the surface from a variety of angles, thus reducing the influence of any single sun glint. However, the user must take care not to make measurements at increasingly oblique angles as reflections can change and these become a more prominent factor as viewing angles begin to increase.

A further challenge exists at very close range and with very straight on measurements. In these circumstances, the camera can end up viewing a reflection of itself, thus resulting in further inaccurate measurements.

In a similar way to a surface’s emissivity, its reflectivity is determined by the surface roughness and morphology.

Reflectivity (R) is related to emissivity (ε) by R=1-ε so the importance of reflectivity can be reduced by making measurements of surfaces that have a high level of emissivity, ideally greater than 0.90.

When measuring controlled surfaces like a steel tank on a rooftop, for example, specialized high emissivity/low reflectivity matte flat black paint can be used to create specific ‘measurement patches’ that then result in extremely reproducible measurements.

Effects of Atmosphere

The Earth’s atmosphere both absorbs and emits infrared radiation. It does this in varying amounts based on factors such as air temperature, air density, relative humidity (RH) and the distance between the camera and the object’s surface. The amount of absorption or emission of infrared radiation from the atmosphere has a marked effect on thermal imaging readings, so must be taken into account.

Atmospheric moisture like fog can greatly reduce a thermal camera’s detection range. Also note the thermal reflections of the person and building on the surface of the water.

Figure 4. Atmospheric moisture like fog can greatly reduce a thermal camera’s detection range. Also note the thermal reflections of the person and building on the surface of the water.

Atmospheric transmission between the camera and the surface can alter the radiometric temperature measurement. This is effectively a measure of the heat reaching the camera, and failure to account for this atmospheric extinction will result in measurements that look colder than the actual surface temperature of the object.

For example, on a warm, humid day, a 100 m air path with 35 °C air temperature and 80% RH would have a theoretical transmission of 80%. What this means in practice is that only 80% of the thermal radiative heat from the object’s surface will reach the camera, so the measurement will be inaccurate.

If the atmospheric transmission loss isn’t taken into account, a camera that is reading a 50 °C object with a known emissivity of 0.97 will read 47.6 °C – that is a 2.4 °C error caused by the air path alone, even without taking into account various other factors.

The simplest way to alleviate the effects of atmospheric transmission is to try and minimize the actual distance between the surface and the camera. To illustrate using the example above, at 10 m the atmospheric transmission rate is 96% and the radiometric temperature (uncorrected for air path) would be 49.5 °C.

The atmosphere is capable of affecting temperature measurements in other, perhaps unexpected, ways too. For example, thermal imaging measurements would ideally be performed away from atmospheric factors like snow, rain, smoke, dust or other obscurants as these too can reduce atmospheric transmission and subsequently change the surface’s background temperature.

Finally, it is important to remember that radiometric measurements can only report the object’s surface temperature, and this surface temperature can also be affected by strong winds.

Resolution

Radiometric thermal images describing a surface are providing a radiometric temperature measurement for each pixel of the image. With that in mind, a small surface within the image will be hard to measure accurately because the number of pixels describing the dimensions of the object’s surface are diminished and not enough to highlight this accurately.

The spot-size effect is the degradation in the accuracy of measurements which is due to the effects of optical distortion, stray light, diffraction and sensor image processing. These result in a washed-out image if not properly accounted for.

The spot-size effect, caused by optical and physical characteristics of the camera, is most relevant for remote temperature sensing.

Figure 5. The spot-size effect, caused by optical and physical characteristics of the camera, is most relevant for remote temperature sensing.

In fact, not properly accounting for this spot-size effect can provide measurements that are significantly influenced by nearby surfaces; for example, a cold object may appear warmer than the actual temperature or vice versa.

While spot-size effects may be very dependent on the thermal camera itself, measurements conducted in the FLIR laboratory do suggest that a measurement spot in the thermal image should be at least 10 pixels in diameter to report an effective measurement and a 20-pixel diameter is large enough to wholly negate the spot-size effect.

The number of pixels used to resolve the surface area of an object is dependent on the pixel pitch, focal length, distance from the camera to the surface, and the smallest characteristic size (length, diameter) of the surface area.  As the distance between the surface and the camera increases, the spot-size effect becomes more relevant as the number of pixels describing specific spatial features is reduced. Pixel pitch, the smallest characteristic size of the surface, focal length and the

Figure 6. The number of pixels used to resolve the surface area of an object is dependent on the pixel pitch, focal length, distance from the camera to the surface, and the smallest characteristic size (length, diameter) of the surface area.

As the distance between the surface and the camera increases, the spot-size effect becomes more relevant as the number of pixels describing specific spatial features is reduced. Pixel pitch, the smallest characteristic size of the surface, focal length and the distance from the camera to the surface are the key factors in determining the number of pixels used to resolve an object’s surface.

The number of pixels (N) used to resolve an object is evaluated by the ratio of the camera angular subtense and pixel instantaneous field of view (N=α/IFOVp). Where the surface to camera angular subtense α=d⁄s is the ratio of the distance between the camera to object surface (d) and the size of the object (s). The instantaneous field of view (IFOVp) of each camera pixel is calculated by taking the ratio of the pixel pitch (p) and focal length (f), IFOVp=p⁄f.

The relationships between these factors can be adjusted and adapted to work out the maximum recommended measurement distance, camera characteristics and smallest necessary object size for any type of radiometric temperature application.

To illustrate this with a practical example, consider a Tau 2 camera with a 13mm lens, 17 μm pixel pitch and 640x512 pixel resolution sensor. Working on the assumption that the camera is 20 m from its intended target, a 30 cm square image directly below the camera would be made up of 12x12 pixels in a thermal image.

Furthermore, an increased number of pixels may be required to make accurate radiometric measurements where thermal images are out of focus due to target or imager motion. Returning to the example of the Tau 2 camera; this device has a sample time of 1/30 second, meaning that a fast moving target may cause a blurred image and subsequently a less accurate measurement.

If a hot surface is smeared out because it is moving, it will likely appear cooler and a cool surface may appear to be hotter. Active-stabilization techniques and camera mounts can be used to reduce camera jitter, thus maximizing image focus and camera stability.

Table 2. Recommended maximum viewing distance for different target sizes, illustrated for a Tau 2 640/13 mm lens. Image blurring and optical distortion will reduce the recommended distance.

Minimum Surface Dimensions (cm) Maximum Distance for 10 pixel spot size (m)
5 3.8
20 15.3
40 30.6
60 45.9
80 61.2
100 76.5

 

Summary

As can be seen, there are numerous factors that can affect the accuracy of radiometric surface temperature measurements. Surfaces with high reflection and low emissivity benefit from the avoidance of oblique view angles – this can reduce the impact of reflections and exacerbated oblique reflection.

Surface emissivity should be high in order to minimize the impact of sun glints and background temperature reflection. Ideally this should be greater than 0.90. Another way to achieve this is to couple a rough surface texture with high emissivity, using matte flat black paint to cancel out the impact of high reflectivity and uncertain emissivity.

Taking measurements within 10 m or less of the target surface, and in a clear, cool atmospheric setting can sufficiently negate any atmospheric transmission factors such as relative humidity, air particulates and air temperature.

When taking measurements at longer distances is unavoidable, atmospheric conditions such as distance, temperature and humidity must be characterized and taken into account in order to calculate the atmospheric transmission.

Measurements that are taken further than 10 m away can also be impacted by the spot-size effect. This causes the number of pixels describing a particular surface to drop and as a result of this, the ability to make measurements of very small objects is impacted considerably.

While it remains important to make measurements with a spot-size diameter of 10 pixels or more, out of focus or blurred images require more pixels than this to be measured accurately.

References and Further Reading

  1. G. Gaussorgues and S. Chomet, Infrared Thermography. Springer, 1994, page 47.
  2. M. Modest, Radiative Heat Transfer. Elsevier Academic Press, 3rd Edition.

This information has been sourced, reviewed and adapted from materials provided by FLIR Cores and Components Group.

For more information on this source, please visit FLIR Cores and Components Group.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    FLIR. (2019, October 21). Different Factors Which Impact Thermal Imaging Quality. AZoSensors. Retrieved on December 10, 2019 from https://www.azosensors.com/article.aspx?ArticleID=1156.

  • MLA

    FLIR. "Different Factors Which Impact Thermal Imaging Quality". AZoSensors. 10 December 2019. <https://www.azosensors.com/article.aspx?ArticleID=1156>.

  • Chicago

    FLIR. "Different Factors Which Impact Thermal Imaging Quality". AZoSensors. https://www.azosensors.com/article.aspx?ArticleID=1156. (accessed December 10, 2019).

  • Harvard

    FLIR. 2019. Different Factors Which Impact Thermal Imaging Quality. AZoSensors, viewed 10 December 2019, https://www.azosensors.com/article.aspx?ArticleID=1156.

Ask A Question

Do you have a question you'd like to ask regarding this article?

Leave your feedback
Submit