Attachment 17 NVImageJ Instructions.pdf

NVImage J Instructions

The NVImageJ software program (Win32 and Win64 versions) is provided to support Offeror's SITF, 3D

Noise, and Uniformity calculations. Once installed, the program can be used to compute SITF, 3D Noise, and Uniformity calculations as shown below.

From the plugins tab, select the 3DNoisePro program and then the CAL 3DN program.

Figure 1: The Tests Tab from the Calculate 3D Noise program.

Select the SITF, 3D Noise, and Uniformity calculations. Move on to the data points tab.

Figure 2: The Data Points Tab from the Calculate 3D Noise program.

Select the “…” next to the Input directory. Point to the folder where all of the images that you wish to process reside.

NOTE: The easiest manner to use the calculation program is to already have the files in multiple frames that are given the filename “BLA_xxdT_BLA.raw”, where the xx is the temperature at which the blackbody was set to for each measurement. Even though the filenames need the dT in them to be automatically have their temperature read, the temperatures can be either absolute or differential.

What is typically easiest to do is have all of the files that you want to use for a calculation in a single folder (for simplicity in selecting imagery and for reasons explained later in this document)

The number of frames in each file should be around 100 (between 64 and 128 are typical for this measurement, and NVESD collects 128 for all noise measurements). Since you are importing raw files, you will then be given the Import screen for the data. Input the information as it pertains to all of your imagery (obviously it makes sense that all data is in the same format for your imagery).

Figure 3: The Import Tab from the Calculate 3D Noise program.

Once you fill out the correct information and click ok, the data points tab will auto populate with all of the information for filename and delta T. Do not worry about the cal factor column yet as it will come from the SITF measurement. All of the files that you originally selected are also sorted by delta T, or at least should be if everything was inputted properly.

If all of the temperatures are correct and the correct files are selected, push the “Create Stack” button at the bottom of the screen. It will take the average of all of the files for each temperature, and then put that average frame into a new stack which is the same number of frames as the total number of input images that you chose.

When sorting through your stack, the imagery should be increasing in brightness from the first frame to the last frame. Please quickly go through this step to make sure that all data is being inputted into the program properly and is correctly formatted for the program. DO NOT CLOSE THE STACK

Once the stack is verified move on to the Tgt and Bk Properties Tab.

Figure 4: The Tgt and Bk Properties Tab from the Calculate 3D Noise program.

From the main ImageJ Program, the Rectangular selection tool should be depressed by default, and select the area on the image stack that you wish to use for the 3D Noise and SITF calculation. When

NVESD performs this measurement, the first 4 rows and columns will be ignored, but the rest of the image will be used for SITF and 3D Noise calculations.

This measurement will likely be performed on digital NUC corrected Bad Pixel replaced imagery that was taken by looking at a flooded field BB.

Once the selection is completed, move on to the SITF Tab.

Figure 5: The SITF Tab from the Calculate 3D Noise program.

NVESD engineers use a third order fit to the data, and the collimator transmission will be 1.0 if the sensor was placed close to the aperture of the BB. Click on the 3D Noise tab.

Figure 6: The 3D Noise Tab from the Calculate 3D Noise program.

Save intermediate imagery is not necessary for non-NVESD users except when some images have unknown values that seem incorrectly high.

For remove dead pixels, NVESD engineers ignore pixel values that are more than 3 standard deviations from the mean pixel value in noise calculations.

For trend removal, NVESD engineers use a blur kernel of size 8.

For fit sigma v and sigma h, NVESD engineers use a fit order of 3 for the data.

Move on to the Uniformity Calculation

Figure 7: The Uniformity Tab from the Calculate 3D Noise program

NVESD engineers will use a blur kernel size of 8 for uniformity calculations.

Finally, go to the calculation tab and click run calculations.

Interpreting Output

Figure 8: SITF of the original data

Figure 9: SITF of the original data with the fit on top of it

Figure 10: All of the 3D noise values as a function of temperature

Figure 11: Results data for the calculation. Temp is the adjusted temperature. Avg Tgt is the average

TGT value. Avg Bg is the average BG value (which will all be zero since these are flooded field calculations), Avg SiTF Tg-Bk is the difference between the target and the background, and SITF fits is the polynomial fit to the SITF as a function of temperature. Fit errors is the difference between the fit and the raw average data, fit Dev is the derivative of the SITF Fits function at each of the temperatures, and cal factor is 1/Fit Dev.

All of the noise values are calculated with dead pixels removed (in this case 3 standard deviations from the mean pixels are thrown out), and the PercentGoodPixels is the percentage of pixels (up to 1) that are used in each calculation.

NETD is simply the standard deviation of the image cube divided by the calibration factor at each temperature.

The next columns have the data with all pixels included in the calculation . It is not necessary, but the total number of “good” pixels in this calculation will always be 1.

Figure 12: The next columns include the image uniformity data for the image cube, where the data here is simply the max pixel minus the min pixel in each region of interest divided by the overall Fit Derivative

(Fit Dev).

Note that this is the format for the sixteenth uniformity calculations.

R1C1 R1C2 R1C3 R1C4

R2C1 R2C2 R2C3 R2C4

R3C1 R3C2 R3C3 R3C4

R4C1 R4C2 R4C3 R4C4

Fit Parameters is the characteristic equation that describes the SITF of the sensor under test. It is in the form A+B*T+C*T^2+D*T^3 for a third order polynomial fit.

Finally, the R^2 fit value is given for the fit to the raw data.