Imaging radiometry data analysis procedure:
This page describes the data processing used in our four-color imaging radiometry method to measure temperature distributions in laser heated diamond anvil cell samples. A description of the experimental aspects of the method, and its performance, is found in Campbell, RSI 2008.
Overview:
We collect data with the SBIG CCD camera, converting output to TIFF files. Then we use the program ImageJ for some image math (subtracting backgrounds and dividing by the calibration image), and export of the results to a text image file. The calculation of temperatures is done in an Excel spreadsheet. In this spreadsheet, the user adjusts the relative positions of four selected regions ("subframes") that each include a narrow-bandpass image of the laser heated spot. When these subframes are coaligned, then the spreadsheet produces a temperature map by fitting intensity vs. wavelength at each position within the laser heated spot.
Obtain Calibration images:
Collect an image of a pinhole illuminated by the calibration lamp. Also collect a background image (blocking light from objective lens, but keeping lamp on). We use a 150 µm pinhole; this is large enough to avoid Airy rings in laser heated region, but small enough not to overlap in the frame.
(Note to UMd users: Normally you will just use the latest calibration file, in the /SBIG camera data/Calibrations/ folder.)
Prepare Calibration files:
Subtract the background from the Calibration image. In ImageJ, this can be done by Process -> Image Calculator. Then use the ImageJ Selection + Analyze->Histogram tool to determine the flatfield background level in image, away from 150 µm spots. (This is an additional background subtraction, removing the flatfield that results from light scattering in the box.) Subtract this flatfield value using Process->Math. Save result as the Calibration file to be used in subsequent data processing.
Obtain Sample images:
Capture an image (at least one, normally many) during laser heating. Either save it as a TIFF file, or convert it to TIFF format (this can be done in the CCDOps program that controls the SBIG camera). Be sure also to collect at least one background image for each set of experiments; normally we do this with the laser on but at a power too low to generate any thermal emission.
Calibrate Sample images:
Subtract the background image from the Sample image, and also subtract a flatfield value using Analyze->Histogram and Process->Math, just as was done for the Calibration file above. Calibrate the Sample image: Divide image by calibration file using Process -> Image Calculator. Store this result as a new file, 32-bit.
Select portion of calibrated image:
The Excel spreadsheet will be used to calculate temperatures, but it requires that we select only a portion of the calibrated image. Use the rectangle selection tool in ImageJ to do this. With Edit -> Selection -> Specify, you must set width=765, height=100, X position = 0, Y position = 200 or something close to it. It is nice to adjust Y so the laser heated spots are centered in the selection. Then crop this selection, and rotate the image 90 degrees to the right using Image -> Rotate. Store this result (should be a vertical strip measuring 100 x 765) using Save As -> Text Image.
Prepare the Excel spreadsheet:
Open a previous Tmap analysis file, such as this one. Open the sample's text image in Excel, copy all of the data in that file, and paste it into the image worksheet (this is found all the way to the right end of the worksheet tabs). Save with a new filename.
Align the four narrowband images of the laser heated spot:
The select spots worksheet is where the 4 narrowband images will be spatially aligned. This worksheet contains 4 separate subframes of the image sheet, one each for 670 nm, 750 nm, 800 nm, and 900 nm. Below these 4 subframes are two graphs that show profiles vertically and horizontally across the 4 subframes. The object is to adjust the positions of the subframes to optimize the superposition of the profiles shown in the two graphs below. The cell numbers that are called by the subframes are in bold blue; these blue values are the adjustable parameters that the experimenter must use to align the subframes. First, adjust the 750 nm frame (row and column) until it is centered around the 16th row, 16th column in the plots below. (Each subframe is 31 x 31 cells, so the 16th is the center.) The 750 nm profiles are plotted on the left-hand Y-axis, which should autoscale for convenience. Then adjust the row and column values (blue cells) for the 670 nm subframe. These profiles are plotted on the right-hand Y-axes, and may require rescaling to help you see the closeness of alignment. Note also that the 670 nm subframe allows not only integer adjustments of row and column position, but also decimal values. These decimal values (range 0.0 to 1.0 only) are interpolations between neighboring rows/columns in the 670 nm subframe. For best results, use these interpolations. The object is to match the positions in the X direction; the Y scale of the plots doesn't matter, you only adjust the Y scale to improve your ability to make the X alignments. Note also that you may need to iterate row/column adjustments a couple of times. When the 670 nm profiles are as close as they can be to the 750 nm profiles, then repeat the procedure with 800 nm and 900 nm. In each case, you leave the 750 nm row/column positions fixed and make adjustments to the 670/800/900 nm subframes to overlay them onto the 750 nm subframe.
Check results:
When the alignment procedure is complete, examine the T map worksheet. The figure on the left is the temperature map of the laser heated spot. The figure to the lower right is another perspective of the T map, and the upper right figure is a T vs. emissivity plot across the T map. Also in this worksheet you will see: a 31 x 31 set of cells with the temperature values in them; a 31 x 31 set of cells with emissivity values; and a 31 x 31 set of uncertainties on the T values (calculated from the misfit to the Wien approximation to the Planck function). There are two blue (user adjustable) entries in the upper left part of the worksheet. The cutoff intensity value adjust the level below which the temperatures are ignored; this is used to clean away the low-intensity stuff that isn't useful in the plot. The row value selects the row number of the T vs. emissivity data used in the plot at right.
Some details:
In the spreadsheet, the calculation proceeds across worksheets from right to left, beginning with the image tab and finishing at the T map tab. The polynomial coefficients for the calibration lamp are at the top of the polynomial coeff. worksheet. The calculation uses the Wien approximation to the Planck function; this is in the ln (I L5) tab. In the ordered tab there is a graph showing the Wien plots for various points across the laser heated spot; these should be close to linear, at least for the useful, high-intensity data. The next two tabs contain some bookkeeping and calculation of T and emissivity from the slope and intercepts of the Wien fits, and the final (leftmost) tab shows the final results.
I look forward to improvements to this procedure. If you make some, please share them! Meanwhile, if there are any questions, email me.
