1 Description
[hide]
In color cameras, it is increasingly popular to use dual-band filters such as the Optolong L-Ultimate, Askar ColorMagic series, and similar devices. However, extracting the specific signal from each emission line (SII, Hα, OIII and Hβ) from OSC data is not straightforward.
DBXtract computes per-channel contributions by taking into account the quantum efficiency (QE) of different Sony IMX sensors and the approximate transmission of each emission line in the RGB channels. The script works on linear images and separates the emission lines into independent grayscale images that can later be combined into SHO, HOO, HSO or mixed palettes.
For example, the following graph represents the QE of the IMX571 sensor. By tracing vertical lines at the wavelengths of the main emission lines captured by common dual-band filters, we can estimate the relative contribution of each line to the R, G and B channels. This means that it is not correct to assign Ha or SII directly to the R channel, nor OIII to the G or B channels, without compensating for these overlaps and for the background pedestal of each channel.
Mathematical overview
Conceptually, we can write the RGB signal of a dual-band image as a linear combination of the emission lines:
R = aSII*SII + aHa*Ha + aOIII*OIII + aHb*Hb + pR G = bSII*SII + bHa*Ha + bOIII*OIII + bHb*Hb + pG B = cSII*SII + cHa*Ha + cOIII*OIII + cHb*Hb + pB
where the coefficients aSII, aHa, aOIII, aHb, etc. depend on the sensor QE and filter transmission, and pR, pG, pB represent the background pedestal of each channel. For a given sensor, DBXtract precomputes these coefficients and estimates the pedestals directly from the image using robust statistics (mode-based background measurement).
In matrix form, for the vector of RGB values C = (R, G, B) and the vector of emission lines L = (SII, Ha, OIII, Hb), we can write:
C = A * L + P
where A is a 3x4 matrix of contributions and P is the RGB pedestal vector. For a given dual-band filter configuration (HO, SO, SHb), the script inverts the corresponding linear system and solves for each emission line. In practice, this is implemented through PixelMath expressions that:
- subtract the estimated pedestal of each channel; - apply the sensor-dependent coefficients to isolate SII, Ha, OIII and Hb; - enforce reasonable limits to avoid over-subtraction in low-signal regions.
When both HO and SO (and optionally SHb) masters are available, OIII and SII can be combined by comparing their statistics in each image. The Integration Method parameter controls how these masters are merged (adaptive scaling, SNR-weighted average, simple average, minimum or maximum), always working on linear data.
Additionally, DBXtract exposes the effective RGB formula used to map the extracted emission lines into a color image. For a given sensor, the script provides default coefficients for:
R = rOIII*OIII + rHa*Ha + rSII*SII + rHb*Hb G = gOIII*OIII + gHa*Ha + gSII*SII + gHb*Hb B = bOIII*OIII + bHa*Ha + bSII*SII + bHb*Hb
and allows the user to customize them in the Sensor QE to RGB Formula section to experiment with different palettes while keeping a physically motivated starting point.
2 Parameters
[hide]
2.1 Camera Sensor
Allows selecting the most popular sensors used in color astrograph cameras. The selected sensor determines the QE-based coefficients used to separate the emission lines and the default RGB mapping coefficients.
2.2 Ha + OIII filter
Selects the view containing the Ha + OIII (HO) dual-band master. This image is used to extract Hα and OIII when only one dual-band filter is available, and as part of the OIII combination when an SO image is also provided.
2.3 SII + OIII filter
Selects the view containing the SII + OIII (SO) dual-band master. This image is used to extract SII and OIII, and can be combined with the HO OIII master according to the selected Integration Method.
2.4 SII + Hβ filter
Selects the view containing the SII + Hβ (SHb) dual-band master, when available. From this image the script extracts SII and Hβ, and can combine SII from SO and SHb into a single SII master using the same integration strategy as for OIII.
2.5 Integration Method
Selects how HO/SO (and SHb, when applicable) masters are combined for OIII and SII.
Adaptive with scaling: normalizes each master by its mean and sigma and combines them in a common scale. Recommended when masters have different background levels, exposure times or calibration.
SNR-weighted average: approximate SNR-based weighted average, using the relative mean levels of each master.
Average: simple mean of the compatible masters.
Minimum: pixel-wise minimum. Useful to reject gradients or residual noise.
Maximum: pixel-wise maximum. Useful to emphasize faint structures present in only one of the masters.
2.6 Customize formula (Sensor QE to RGB Formula)
Enables manual editing of the RGB combination used to build the color image from the extracted SII, Hα, OIII and Hβ channels.
By default, the coefficients are derived from the QE model of the selected sensor. When this option is checked, the user can override them to define custom SHO, HOO or hybrid palettes.
