TRICOLOR PROCESSING TUTORIAL

Top, the red part of the spectrum (TM57-1 image). Bottom, the green part of the spectrum (TM57-2 image). This images are captured in difficult sub-urbain conditions (naked eye magnitude of 2.0) with the T190 mm and Audine camera.

We first isolate the following monochromatic images with Iris: H-alpha (at 6563 A), [ NII ] (at 6584 A), [ OIII ] (at 5007 A) and HeII (at 4686 A).

One notes on the spectrum that the line H-alpha is covered partially by the line [NII ] at 6548 A. It is a serious problem! We will use an easy way for resolve this problem: we subtract the 6584 A monochromatic image to the 6548 A monochromatic image. We suppose here that the 2two lines [ NII ] are homothetical images of M57.

Portion of the starting image TM57-1.

The consecutive steps of the processing are:

Translate towards the left so that 6584 A image superimposes the line 6548 A image. The value good of the translation (-38.3 pixels) was determined by successive trials:

TRANS -38.3 0

Multiplication of the shited image by a constant value so that 6584 A and 6548 A images have the same intensity (determined by
successive trials):

MULT 0.34

Back up the result:

SAVE I

Subtraction of the shifted image with the starting image:

SUB I 0

Normalise the sky background to an arbitrary value of 500 ADU (Analog Digital Unit, or DN for Digital Number):

OFFSET -255

Save the result on disk:

SAVE R

The line H-alpha image is now clean.

Extact the line [ NII ] at 6584 A and enlarge the image:

WINDOW 392 152 414 196
SCALE 3 4 4
WINDOW 1 1 80 172
SAVE I6584

Isolate the line H-alpha and enlarge the image:

TRANS 22.3 0
WINDOW 392 152 414 196
SCALE 3 4 4
WINDOW 1 1 80 172
SAVE I6563

We now will extract the lines [ OIII ] at 5007 A and He II at 4686 A in the M57-2 image.

Taking into account the optical spectral transmission, spectral efficiency of the grating and the CCD spectral responsivity, we consider the following ratio for the relative sensiblity versus wavelength:

 Wavelength (A) Relative sensitivity of the total instrument 6563 1.00 5007 1.06 4686 1.03

We begin by adjusting the intensity of the [ OIII ] image in manner has to be coherent with the intensity of  6563 A image. Then, we ajust the sky level to 500 ADU (the same one as for H-alpha and [ NII ]):

MULT 0.94     (0.94 = 1/1.06)
OFFSET -300

We complete the image so that it have the same size as M57-1 (note: the INFO command return size of the current image):

We register the [ OIII ] image so that it is superimposed on the line [ NII ]:

TRANS 12.1 92.1
SAVE I

Checking:

SUB I 7000
VISU 12000 0

We extract the line [ OIII ] at 5007 A:

WINDOW 392 152 414 196
SCALE 3 4 4
WINDOW 1 1 80 172
SAVE I5007

Finally we get the HeII line at 4686 A:

MULT 1.03
OFFSET -43
TRANS 342 0
WINDOW 392 152 414 196
SCALE 3 4 4
WINDOW 1 1 80 172
SAVE I4686

From left to right the monochromatic images of M57 at 6584A, 6563A, 5007A and 4686A.

A tri-color image synthetised from the monochromatic images of M57. The R-chanel correspond to the [ NII ] 6584 A images. The G-chanel to the H-alpha 6563 A images. The B-chanel correspond to the [ OIII] 5007 A.

One notes at once that there is very little green in this image. This shows that the H-Alpha line contributes little to the colored composition, because it is relatively weak.

In the images above it is evident that the M57's images are abnormally lengthened stretched vertically. The cause is the anamorphic propriety of our spectrograh. The vertical axis is stretched by a factor 0.80 compared to the horizontal axis in the R=3000 spectrograph. I used again the function SCALE to correct this problem but also to stack the monochromatic images to an high resolution wide band KAF-0400 image of this nebula (panchromatic image, intrument: Celestron 11) extracted from the Buil-Thouvenot Atlas. Then, we exploit LRGB capability of Iris. For example:

LRGB I6584 I5007 I4686 PANCHRO R G B
TRICHRO R G B

Left, the standard tricolor image. Center, the panchromatic image or grey image. Rigth, we use LRGB command of Iris to make image fusion and so a high resolution color image (the I image in the HSI domain is replaced by the panchromatic image). R-chanel=[NII], G-chanel=H-alpha, B-chanel=[OIII].

For the left image we use the arangement R = [NII], G = H-alpha, B = He II (the intensity of the He II 4686A image was multiplied by 40 here relative to the other lines). This image confirms that the line H-alpha contributes little to the brightness of the nebula. For the right image we adopt this configuration R = [NII], G = [OIII], B = He II. Note that differences between [NII] and [OIII] are notably significant and that the He II line emit only in the central part of the nebula.

Left, the variation of the [NII] to H-alpha intensity ratio. The ratio varies from 0.32 to 2.51. Right, the isophotal contour of the intensity ratio map with start/end/step=0.32/2.51/0.1.

The Iris software include Pincipal Component Analysis (PCA) methods. The PCA is a very powerful mathematical tool useful for the study of the very weakly colored images, i.e. highly correlated images. The principle of the PCA is to obtain a small number of uncorrelated images in which the differences of colours are very hardly stressed. Try for example the RGB2PCA command:

RGB2PCA I6584 I6563 I5007 C1 C2 C3
TRICHRO C1 C2 C3

The first principal component image (left) accounts for as much of the variability in the colored image as possible (R-band=[NII], G-band=H-alpha, B-band=[OIII]. Each succeeding component (mid, rigth) accounts for as much of the remaining variability as possible.

Iris color display.