The principle: an introduction
For temperatures around room temperature, for
example, the operator will set the camera to a typical temperature range
of -20 °C to +50 °C. All objects with a temperature beyond this range,
the brightest or hottest parts, will be presented as saturated. Those
below the range will generally be noisy.
So, if for instance our object temperature of interest is +100 °C, a range of +20 °C to +120 °C must be selected. In this case the camera will present a good image of the +100 °C object, but the detailed contrast of the room temperature objects within this image will not be as good as displayed in the first, - 20 °C to +50 °C range.
The combination of the two images would be a
logical step. The solution would be to allow the camera to “take” an
image in the first, room temperature range and then “take” a second
image at the higher temperature range. Combining these two images in a
smart way should produce a superior picture which would encompass the
best parts of both images. This is what Superframing is about.
 |
 |
 |
 |
|||
| Tint = 1.0 ms Range = 20C -75C |
Tint = 0.25 ms Range = 65C -135C |
Tint = 0.05 ms Range=130C-230C |
Tint = 0.01 ms Range=220C-380C |
Pictures 1 to 4: Welding, a sequence with extreme temperatures: the shorter the integration or exposure time, the higher the temperature range and the larger the black or noisy areas.
The problem and the application
The matter gets more complicated when dealing with extreme temperatures: a man standing by a fire in a cold winter night is such a typical example. The brightest or hottest parts of the image will be saturated, while at the same time the darkest or coldest parts of the scene will appear black in the image, or noisy. When an object appears saturated or noisy, two undesirable things happen: image details are lost and temperature measurements in that part of the scene are no longer valid.
Advanced thermal imaging and measurements often have to acquire images or image sequences of scenes featuring a very wide range of temperatures. The saturation problem can be particularly disturbing in thermal imaging for research and development applications, where imaging or high-speed digital video of scenes with very large temperature differences such as engine monitoring, a rocket launch or an explosion are required. This problem, which is particularly acute in the Midwave Infrared waveband, can be addressed with Superframing.
Will reducing exposure (integration) time do it?
The sensitivity of an thermal imaging camera,
which means the smallest temperature difference such a camera can
detect, can be controlled by varying its exposure or, as we call it for
thermal imaging systems, integration time, a term which we define as the
exposure time of the thermal imaging detector inside the camera to
produce a single frame.
Operating the thermal imaging camera at a
longer exposure time raises sensitivity, but at the same time it
restricts the range of temperatures one can measure: the hot objects are
so bright that they exceed the set temperature range of the camera. If a
scene or sequence contains extreme temperature differences which need
to be simultaneously measured, the exposure time of the camera should be
reduced to a much shorter time. But this reduction, for its side, will
result in the loss of the ability to measure changes in the cooler parts
of the scene as it exceeds the set temperature range, resulting in a
black or noisy appearance for those areas, as shown in the picture range
1 to 4.
Is there a single exposure time able to fully
encompass a scene’s variations and accurately measure accurately all
objects, cold and hot, in such a scene? No, but there’s another option.
The solution: FLIR Superframing
FLIR Superframing means taking a set of typically four images (subframes) of the scene at progressively shorter exposure times in very rapid succession, then repeating this cycle. The subframes from each cycle are merged into a single superframe which combines the best features of the four subframes, which as we already know, have different exposure times. This process is called collapsing. In this way, the superframe image, generated by the collapsing algorithm, will be both high in contrast and wide in temperature range. The algorithm is basically quite simple: if a pixel in the first subframe is saturated, the algorithm selects the corresponding pixel from the next subframe. If that pixel is satisfactory, the algorithm stops. If not, it checks the corresponding pixel in the next subframe for suitability, and so on. All pixel values are converted into temperature or radiance units for the final superframe image.
 |
 |
 |
||
Figure 1: 2-millisecond image: saturated exhaust system |
Figure 2: the 30-microsecond image shows the exhaust system very clearly without saturation, but the rest of the scene is too cold to see clearly above the system noise floor |
Figure 3: The solution: a picture both high in contrast and wide in temperature range |
Figures 1, 2 and 3 illustrate the Superframing
technique with two images of a Beechcraft King Air twin propeller
airplane taken at 2 milliseconds and 30 microseconds. These images were
taken with a FLIR SC7000, a very high performance midwave infrared
(MWIR) camera system, running at 170 frames per second at the full frame
size of 640x512 pixels. The two images are separated by a short
interval of time (about 40 milliseconds), meaning that the scene does
not change very much; the propeller movement is barely perceptible.
The 2-millisecond image gives excellent contrast for nearly every portion of the scene except for the aircraft’s exhaust system, which is so hot that it saturates that part of the image. (Fig. 1) Conversely, the 30-microsecond image shows the exhaust system very clearly without saturation, but the rest of the scene is too cold to see clearly above the system noise floor. (Fig. 2) Combining these two images with the right algorithm enables imagery that is both high in contrast and wide in temperature range. (Fig. 3)
Technology
There are some technological preconditions to
apply Superframing which luckily have emerged in the commercial
marketplace. One is the emergence of commercially available thermal
imaging cameras with large arrays such as 320x256 or 640x512 pixels that
are able to produce high frame rates which are necessary to produce
superframes. The other is the availability of computers able to process
the enormous amounts of data (64 MB/s) coming from the thermal imaging
camera when operating at high frame rates.
This technique has already been implemented in
the R&D and Science advanced solutions of the FLIR SC Line. All the FLIR SC2500 - SC5000 - SC7000 Series and its ALTAÏR image processing
software, offer an integrated FLIR Superframing function.
Summary
FLIR Superframing dramatically extends the
effective scene brightness of a thermal imaging system while maintaining
thermal contrast, even at low temperatures.
The technique of FLIR Superframing consists of varying the exposure, or integration time of the camera from frame to frame in a cyclic manner and combining the resulting subframes into single superframes with greatly extended temperature ranges, which allow to visualize scenes featuring extreme temperature differences.
