Missing pixels after Bayer decoding using ASCOM driver

Bernard91

I just started using the ASI224MC color camera (1304x976 pixels) with the ASCOM driver (version 6.5) called from a python code. After exposure, I do get a Bayer-encoded RGGB image from the ImageArray property. However, I was expecting to get an image with four times as many pixels due to the extra space needed for the three color channels. Instead, the array I get is only 1304x976 big, so that after Bayer decoding the final color image has only 652x488 pixels. How can I get the missing pixels ? Did I misunderstand the method required to get a color image or is there a problem with the ASCOM driver in case of a color camera ? I checked that SensorType is equal to 2.

w7ay

Bernard91 However, I was expecting to get an image with four times as many pixels due to the extra space needed for the three color channels.

I think you misunderstand how a Bayer Array works.

A Bayer Color camera is simply a monochrome camera with small tiny R G and B filters placed over each pixel. The filters are sometimes CMY, sometimes with a luminance ("white") channel, etc.

The number of pixels don't change, but each pixel responds to different passbands in the spectrum.

The cameras that ZWO sells have RGB filters, and arranged in a 2x2 supercell that looks like this pattern repeated over the sensor:

RG
GB

It is usually abbreviated RGGB.

The deBayering process then attempts to reconstruct a color image (by lowpass and resampling, usually) from that. After deBayering reconstruction, the result has the same number of pixels as the monochrome sensor, but each pixel now has all (degraded) color components.

See

https://en.wikipedia.org/wiki/Bayer_filter
https://patents.google.com/patent/US3971065A

Chen

Bernard91

You're absolutely right, I misunderstood. Although I got Bayer filtering right, I thought that my camera had 1304x976 supercells of 2x2 RGGB pixels, when you tell me that it is actually 4 times less. Indeed, the link you provide explains clearly that deBayering algorithms provide a final image that is actually oversampled with respect to the true physical resolution of the camera.
Thanks a lot for your explanation!