转发: RGB Stack

[Donglai Gao]

Dear sirs and madams,
My name is Donglai GAO, and I am a fresh user of ImageJ.
Recently, I am using ImageJ to process images. Basically, it can be described as a background subtraction. I found Imagej could generate the background by simply using the 'Image-type-RGB Stack' function. Please refer to the attachment, Fig. 1 is the raw image and Fig.2 is the background converted with ImageJ. By a subtraction, the desired red rivulet is obtained, as shown in figure 3.
However, I don't understand the algorithm of the convert process incorporated in ImageJ. Would you please give me some instructions or reference papers? Because I am trying to explain this in a mathematical way. Thank you so much.

Sincerely,
Donglai

----------------
Donglai Gao, Ph.D. Candidate
IRPHE, École Centrale Marseille, France

[cid:8c91ccc8-684a-4fdf-b515-20860be3f97e][cid:bfad86be-533c-40ae-bac5-11c5c21ef477][cid:e9c3c2c7-4169-4c4c-87f3-5c5265a9d3e3]



--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Michael Schmid]

Hi Donglai,

as I understand it, what you are doing is the following:
With Image>Type>RGB Stack you get the three color channels, with the 'R'
channel being the first one. You then see the red channel (you could see
the other ones with moving the slider at the bottom or the '.' and ','
keys, but you don't use them).
For the red features in the image, the red intensity (pixel value) is
the highest; for everything else the pixel values of the other channels
are comparable to or more than the red channel.
When you transform to grayscale (this automatically happens when
subtracting a color image from a grayscale image like the 'red'
channel), ImageJ calculates a (weighted) average of the RGB channels of
the original:
   https://imagej.nih.gov/ij/docs/guide/146-28.html#sub:Type
   https://imagej.nih.gov/ij/docs/guide/146-27.html#sub:Conversions...
So you subtract the (weighted) average of the other color channels from
the red channel. Compared to other colors in the image, this will give
you the highest result for red features in the image.

By the way, try to avoid using JPG images, rather use uncompressed
images (if your camera can create uncompressed images such as TIFF or
RAW). For operations with color channels, the artifacts of JPG
compression can be quite severe (you may notice strange block-like
features after the subtraction, these are due to JPG compression).
Of course, if you have created JPGs only for the mailing list (to keep
the file size small), this is perfectly ok.

Michael
________________________________________________________________
On 31/05/2017 12:31, Donglai Gao wrote:

> Dear sirs and madams,
> My name is Donglai GAO, and I am a fresh user of ImageJ.
> Recently, I am using ImageJ to process images. Basically, it can be described as a background subtraction. I found Imagej could generate the background by simply using the 'Image-type-RGB Stack' function. Please refer to the attachment, Fig. 1 is the raw image and Fig.2 is the background converted with ImageJ. By a subtraction, the desired red rivulet is obtained, as shown in figure 3.
> However, I don't understand the algorithm of the convert process incorporated in ImageJ. Would you please give me some instructions or reference papers? Because I am trying to explain this in a mathematical way. Thank you so much.
>
> Sincerely,
> Donglai
>
> ----------------
> Donglai Gao, Ph.D. Candidate
> IRPHE, École Centrale Marseille, France

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html