|Operator Instructions||Source code|
Ronchical applies the camera corrections (dark, clamp and smear) and averages all 6 bands. It then uses the average calibration image ((east + west)/2) and the diffuser image to generate a flat-field that is divided into the ronchi image to remove high spatial frequency camera artifacts.
At this point, if one plots a line across the image, we see a periodic signal, but it more closely resembles a sin**2 function than it does simple sine function.
Therefore, we take the square root, giving:
Which is much more "sine wave like".
Now we need to compute a constant that, when subtracted from the ronchi image, will produce a sine wave that is symmetric about zero. Since the distribution of data values in the ronchi images is essentially bi-modal, a simple mean or median subtraction does not work. Instead, we compute a histogram of all of data values within a specified distance from the center of the image. Using this we find the "central-value".
An example of how the central value is computed is shown in this figure: (Ronchi Histogram plot)
Finally, the image is normalized such that the peak-to-peak values are -1.0 to 1.0 .