The purpose of this article is to review different ways of processing a color image of Mars, and to see how those different methods affect the restitution of the information. The idea here is to say that the RGB method is preferable to the LRGB.

Let’s review what kind of information we are expecting to see in each given colour :

RED in the red are seen the details of the surface, dark markings, light deserts, and the polar caps. Dust storms are usually brighter in red also.

BLUE the blue shows the information of the martian atmosphere (white clouds).

GREEN while it is often stated that green doesn’t contain any useful information for Mars, maybe it’s time to re-examine our judgment on this. The green, as it shows also the surface’s details, thus contains a part of the information that forms the colour and the shape of those details, in a way that may not be fulfilled with the R(G)B method (see below). Furthermore, they are actually a kind of information that could be best seen in green than in red : the light dust hazes or faint dust clouds (not thick dust storms) are likely to be best seen in this colour, as the red (or IR) may penetrate a slightly dusty atmosphere.

We can start from the the following point : in the visible range, the information that forms a detail is distributed in the three colors (or at least two). For example, the information about dust activity could be formed by, let’s say, 60 % of red (the core of the dust clouds), 30 % of green (the further spreading of airborne dust outside the main cloud), and 10 % of blue (idem + possible water vapour). So reproducing correctly the whole information would mean combinining the three types of information in their respective weight.

From this starting point some remarks can be made that could lead to say that the LRGB method is not satisfying :

1)To put in luminance a R or IR image leads logically to a change in the distribution of information, giving more importance to the red than to the others two colour, or even to a loss of the information contained in G or B (like faint white clouds).

2) The original LRGB method (with integrated light in luminance) would give a better result, but it’s not possible to do one with webcams (unless one got also a B&W webcam), and even here it could be difficult to reproduce white clouds.

Let’s take a look at those different processing with images taken on august 1^st, 2003 :

RGB : The image is a three-color one realized with a b&w webcam and interferential filters.

RRGB : The image taken through the Red filter is put in luminance above the unprocessed RGB.

IRGB : An image taken through an infrared filter (IR) is used for the luminance.

R(G)B : The RGB is built with a synthetic green image, mean value of R and B.

Here are some advantages of RGB in comparison with the other three methods :

1) Colours (assuming that the RGB above is close to the visual tints)

If colours look good in every image in general, there are some differences in the details. First each LRGB change the generals colours : the dark markings are now greyish and the light deserts are pink instead of orange. This is due to the fact that LRGB is very different from RGB : it colorizes a grey-level image, and this original grey explains why the colours of the chrominance are not fully respected (it should be preferable to fully process also the chrominance). Then much colour shadings are lost in the LRGB images but also in the R(G)B : above all, the yellowish tint of the Hellas basin in the RGB is not reproduced in the following images. This also shows that the synthetic green is not equal to a true green image.

2) Information about the atmosphere

More important, the details about the martian atmosphere are lost or minimized in the LRGB. The above RGB seems to show a vast but subtle white mist in the northern hemisphere (also seen on a ToUcam image taken later with a light blue filter). This subtle veil is totally erased by R or IR luminance. Furthermore, the dark markings at the martian evening look obscured in the RGB by what should be the evening mist (present in the original green and blue). This seems also lost in the LRGB, where the shading is most due to the phase effect than to that veil.

In presence of a dust activity, an RRBG image will also tend to minimize or miss the true extension of dust (which requires the G component to be evaluated fully) ; an IRGB could even minimize the size of the dust clouds themselves.

What do making CCD images mean ?

I would like to cite, in support of this analisis, a reknown Mars observer, OAA'Mars section's Director, Masatsugu Minami, on the effect of LRGB processing in 2001 while the dust was expanding on the planet :

There have been produced a lot of RGB or LRGB images since the start of the dust storm, while we do scarcely come across with the images that may reproduce the realistic tint of 'yellow' intrinsic to the expanding airborne dust cloud. Especially it was strange to us some of LRGB images continued to produce some 'normal' surfaces regardless of the obvious fact that the yellow dust had already proved globally covering the same surfaces (...). Obscuration is a matter of degree, but when we are to discuss how the storm will turn out, we should pay much attention to the subtle obscuration due to the expanding yellowish airborne dust, without depending much on the ability of the infrared penetration. The LRGB image is sometimes powerful, but as is widely known, it has not been well established yet as far as the Martian images are concerned, and we should sometimes refrain from saying something concerning the atmospheric matters if any depends only on the LRGB images. (Masatsugu Minami, 13 july 2001, Director's Report n°10).

These words of M. Minami illustrate perfectly the problem that causes an LRGB processing to who is interested in understanding what's going on on Mars. By changing the repartition of the information, the LRGB image is likely to miss the most important. At this point it can be useful to discuss again what is the meaning of digital imaging, what's it purpose. The most usual judgment here is that the CCD image should reveal "more details than the eye", that it's an excellent tool to reproduce more information. This is exact but let alone, this judgment can lead to careless uses.

It could be possible already to say that the goal of CCD imagery is not only to "show more details" ; the goal is as well to reproduce some information that the eye already perceives, in order to prove that it's real. But the information is accessible to the human eye.

Then, every CCD observer knows that many technics of processing

FURTHER NOTE ON THE R(G)B METHOD

Another processing with images taken on july 23^rd, 2003, shows again what interest can have a true green image :

The RGB at left reveals a light yellow haze over Noachis. The R(G)B image at right with a synthetic green totally fails to shows this haze, which is a very important information. What’s happening is that the information about this haze is contained entirely, or almost, in the original true green image of the RGB : it’s totally impossible to reconstitute this detail with only the information contained in R and B.

_________________________