Debayer

Step: Linear

Turns the raw image of a one-shot-color camera into a color image. Such a sensor records only one color per pixel, in a repeating 2×2 pattern of red, green and blue filters (the Bayer pattern). Debayering fills in the two missing colors at every pixel from their neighbors, so you get a full RGB image. Use it on an opened Bayer FITS file.

Parameters

Parameter	Range	Default	Effect
Bayer pattern	From file header, RGGB, BGGR, GRBG, GBRG	From file header	The color filter layout of the sensor. From file header reads it from the FITS header; pick one by hand if the file names none or names the wrong one.

The pattern is named by the 2×2 cell at the top-left pixel of the image as Siderea shows it: RGGB means red at the top left, green to its right and below it, and blue diagonally opposite. Siderea already accounts for FITS files storing their rows bottom-up, so the pattern from the header needs no correction.

How it works

Every pixel keeps the value it measured for its own color. Each missing color is the plain average of the pixels of that color in the 3×3 neighborhood (bilinear interpolation):

green at a red or blue pixel   = mean of the 4 direct neighbors
red at a blue pixel            = mean of the 4 diagonal neighbors (and vice versa)
red at a green pixel           = mean of the 2 neighbors left and right, or above and below

At the image border only the neighbors that exist are averaged. The result is a 3-channel image of the same size; the capture information of the file is kept, but the Bayer pattern is removed because the image is no mosaic any more.

When to use it

• Only on data that is still a mosaic. Calibrate with darks and flats first, then debayer: dark and flat frames describe the raw sensor, not the interpolated colors.
• If the colors look wrong (red and blue swapped, or a green cast with a checkerboard texture), the pattern is wrong. Undo and try another pattern.
• An image that is already in color cannot be debayered; the module reports an error.