PJ73 Jet N7

2025-07-01 09:45 UT
Credit : NASA / JPL / SwRI / MSSS / Gerald Eichstädt
Submitted By : Maquet-80
Mission Phase : PERIJOVE 73
Source Image(s) : JNCE_2025159_73C00023_V01

PJ73 JunoCam images are very noisy. The noise with mostly positive values seems to be mostly added on top of the underlying signal similar to thermal noise and elevated dark current.

The underlying idea for denoising in this case is hence to use only the darkest pixels within each raw 4x4 image block. More accurately speaking, the pixels of each 4x4 image block are sorted by brightness, and the 3rd darkest brightess tie is assumed to be the denoised value for the whole 4x4 block. Using a darker tie would result in dark artefacts, conceivably due to occasional companding overflow effects of noise peak values.

In order to account for the assumed dark current, an additional value of 48/256 of the raw grey-scale value range is subtracted.

The resulting modified raw image is then used to derive a reprojected RGB image the same way raws are normally used to derive a reprojected image. Nominally, the square-root of linearized values is used for the final image product, with linear radiometric factors (0.613; 1.0; 3.655) for (R;G;B) derived from PJ72.

Note that the applied method is only supposed to reconstruct information about the structure of the cloud-tops. The applied image processing method isn't suited to provide radiometrically meaningfully calibrated values, since decompanding wasn't adapted to the the noise cleaning approach.