It depends very much from relative distances amongst camera, window and target and also from emissivities of taget and grid.

If windows  and target temperature are very similar then you will find many difficulties in any cases.

If target temperature is sensibly different from grid than you can try to characterize the trasmissivity of your IR window placing a black body in the place of the target defining your own calibration curve.

However since we worked a lot in this filed you can contact me any time to explain a bit better your application for more precise suggestion.

thank you very much for your answer. I agree with you according to my experience. I'm also working about this question since some months and I have developed a procedure that does exactly what you suggested. Thus, I'd like to share opinions with someone who has / had these kind of  problems. I will contact you for sure.

Dear Gennaro,

Thermal Imaging through a multiple grids of 1.2 mm dia is not encouraged as the emissivity of the grid material may alter the property of the signal you are interested. To my knowledge, the image could be captured by merging few of the neighboring grids located at the center portion to form a single slot and the remaining portion of the grid needs to covered with a perfect black body material. Thereby, the thermal image could be captured without any interference.

Alternatively, the camera could kept to a closer distance the grid and grab the image; then thermal image processing techniques could be used to segregate the region of interest in the captured image and the same could be analysed.

Mr. Gennaro, I have not worked on this topic but based on the little understanding of your query, the suggestion was given. A request to point out the mis-interpretation of the query raised with limitations, if any.

Regards,

S.Sivanandam

Dear Gennaro,

                          I think there are three issues here.

(1) Emissivity and transmittance of your window/grid. This has to be addressed  by calibration procedure only. You can use standard black body or you can use a black painted specimen with known temperature and can adjust the camera parameters or can obtain a single equation between actual and measured temperature (by camera). A pearson correlation test between the two independent temperature data set will give you good amount of confidence.

(2) If you are only interested in temperature measurement, then the issue ends here. but if you are interested to acquire thermal images, then you need to fall back on image processing techniques like segmentation or image subtraction (object along with grid - only grid). These are speculative suggestions and you need to find out the best technique based on your your application, available resources and time frame. But kindly note that image processing on coloured images are always tricky and may give wrong information. So you may have to work on gray scale images for less complexity.

(3) Finally your experimental geometry must ensure that the field of view is captured by the camera through the window of dia 1.2 mm. This a bit tricky and you need to consider various factors like window to camera distance, manual focusing of camera and most importantly ettendue (range of overlap of cross-section)  of the two solid angles (one from object and one from camera).

As the dia is 1.2 mm and you are in approximately 2.5-5.6 micro-meter wavelength range, you need not worry about diffraction limited optics. 

Dear Umberto, Dear  Barid, dear Sivanandam,

Thank you for your answers. 

With reference to issue (1), I performed calibration. Since the target, the grid and the camera are very close (say within 1 m) and the target applicator-pipe is hotter than the grid (say pipe temp > 40°C while grid is cooled at room temperature), calibration seems to be nice.

With reference to issue (2), I'm using image processing techniques as simple as subtraction. I'm interested to steady conditions.

With reference to issue (3), I agree that diffraction limited optics is not a problem due to the playing wavelength. IFOV problems are compatible with the geometry in use and are encompassed by the calibration.

Since calibration is the critical point, I submit to your expertise the attached text where the fundamental thermography equation (slightly modified for the specific purpose) is shown.

I would be glad If you could have a quick look. I ask for that since it is hard to report equitations in plain text.

I thank you in advance and look forward to hear from you,

Gennaro Cuccurullo.

Dear Gennaro,

                          I have gone through the write up and it seems to be fine. Nevertheless, I would have appreciated if the values of n were discussed. Calculating n is very important as it has a direct effect on the shape of the calibration curve. Otherwise it is fine. I do have one question, how do you separate out f1 and f2 in to two additive functions and not multiplicative. This requires some citations and a few lines of justification. Overall the procedure seems to be correct. Thanks for sharing and best of luck.  

Dear Barid, first of all, thank you so much for your time and for carefully reading my notes. Regarding your observations I can say that you  catched the question: calibration is extremely sensitive to the n-value. This value is inferred from the camera itself, i.e. it is built in the camera inner calibration parameters. Further, I was able to check that the IR termography readout given by the camera companion SW where tthe same I was able to derive by myself when using that value of n. The separation of f1 and f2 into two additive functions arose by observing the calibration curves structure. In order to check the structure I sought, I'm actually performing further experimental tests (with more intense and uniform grid-cooling-flow distribution). Thus, I expect that f1 is more flat than before. Hope to hear from you, more, and thank you again for your relevant observations.