A simple range test was completed today
Before we look at the pictures we need to look at the effect IFOV, as we did previously with the close range tests.
The IFOV of a camera and lens system will effect what can be seen at range and how many pixels are able to see the target at any instant in time. Why is this important ? Well if you have a 1000 pixels seeing the target, that may show a nice thermal response on the display. If 100 can see the target, a lesser, but still useable thermal response will appear in the display. Even at 10 pixels seeing the target the user will be able to see the target. At lower than 10 pixels seeing the target, life can become more complex, especially if there is noise in the image. As a rule the minimum number of pixels you need seeing the target, for it to be discernible, to the user is 4. Less than that and there is a probability that the camera and user will not provide reliable results. The Delta T in a scene also effects the users ability to see the target as a low Delta T scene may mean a narrower span is in use and the noise level in the image can rise, disguising the target. The other posbility is a very large Delta T that forces an auto ranging camera to set a very large span that lessens the contrast in the image.
In this range test I am expecting the camera to struggle due to the combination of its 80 x 60 pixel FPA and and 50 degree wide angle lens. I saw the expected results in my testing. I will detail the IFOV figures below so that readers can see the dimensions of a single pixel 'on target'. If this is unclear.... if the IFOV at 2m results in a pixel size on target of 20mm then a 20mm x 20mm target will illuminate just one pixel. A 40mm x 40mm target will illuminate 4 pixels and a 80mm x 80mm target will illuminate 16 pixels. Bearing in mind my previous comments, the camera will struggle with the target that is less than 40mm x 40mm in size and will not show any shape detail (other than square!) until any least 80mm x 80mm in size.
Multicomp thermal camera IFOV at defined distances:
1m : 11.04mm Recommended Target size for 2 x 2 pixel detection = 22mm x 22mm
2m : 22.08mm Recommended Target size for 2 x 2 pixel detection = 44mm x 44mm
3m : 33.12mm Recommended Target size for 2 x 2 pixel detection = 66mm x 66mm
4m : 44.16mm Recommended Target size for 2 x 2 pixel detection = 88mm x 88mm
5m : 55.19mm Recommended Target size for 2 x 2 pixel detection = 110mm x 110mm
6m : 66.23mm Recommended Target size for 2 x 2 pixel detection = 132mm x 132mm
7m : 77.27mm Recommended Target size for 2 x 2 pixel detection = 154mm x 154mm
8m : 88.31mm Recommended Target size for 2 x 2 pixel detection = 176mm x 176mm
9m : 99.35mm Recommended Target size for 2 x 2 pixel detection = 198mm x198mm
10m : 110.39mm Recommended Target size for 2 x 2 pixel detection = 220mm x 220mm
Now before anybody gets too excited, remember that the above is just for "Detection" and those pixel on target sizes will also dictate how much detail may be seen. That is to say, if a large target contains fine detail, that may not be seen if that details size falls below the threshold of detection. Interpolation of the low resolution image can assist in improving the appearance of an image, but if not enough pixels actually see the detail, interpolation cannot help much, if at all.
The test target that I chose to use in this test is a Leslie's Cube. This is a thermography training device that is a copper hollow cube that has diffrent emissivity surfaces on four of its sides. The copper cube is filled with hot water and the difference in surface emissivity, and its effect on IR measurements, is shown to the students. I am using it as a simple 100mm x 100mm high emissivity thermal target. Its known emitter face dimensions may be used with the IFOV to test the theroy of what I detailed above. For decent detection with the stated IFOV on the Multicomp camera we should image the cubes 100mm x 100mm face at no greater than 4m. We shall see what happens after that 4m point in the images that follow. Remember what I said earlier, we want 4 pixels on the target, with anything less risking issues for the camera and user and there will still be no real detail in the produced target image. In this case we are using a square emitter surface so the shape identification is not demanding ! As we move futher away from the Target I believe we will see the cubes square face lose its edges and become a 'blob' of no defined shape. At a distance from target of 9m, the camera will have only one pixel 'on target' and the camera will likley stugggle to show that clearly to the user. The cameras auto span system may also struggle at this point to set the correct span if the measurement system does not get decent data from the single pixel that sees the target. The targets energy can bridge across two or even more pixels but the pixels will output poor data.
So with all of that out of the way... to the pictures......
Each picture has the distance to target as its name. I produced two test runs. One with center spot measurement, and one without, to give a clear view of the cube to us.
The first pictures are of the Leslie's cube.
Fraser
Home
Help
Search
Login
Register
