RGB Imaging

[jspecs]

Hello,
I am doing a research project on how pumice rocks change color with varying amounts of heat. My findings thus far are that as rocks are exposed to higher temperatures, they oxidize and turn darker and darker shades of red.

I am attempting to quantify my data using Image J. I want to be able to upload my digital images of the rocks into the program, convert my images into RGB images and then determine the mean occurrence of red, green, and blue in each image (potentially with color histograms?). My ultimate goal is to quantitatively represent the color change occurring as the rocks change color with heat by determining the amount of red, green, and blue in each image.

Is this possible with this program? Any help would be much appreciated.

[Michael Schmid]

Hi Julian,

just read the documentation to get a few ideas what could be done:

Image>Color>Split Channels
Analyze>Histogram for each of the channels

or

Image>Type>HSB stack
Analyze>Histogram for each of the stack slices.
Especially the hue slice will be of interest to detect a change of  
hue (see http://en.wikipedia.org/wiki/HSV_color_space if you are not  
familiar with HSB)

Of course, an equally important problem is having reproducible  
conditions when taking the photographs (equal illumination, fixed  
white balance of the camera).

Michael
________________________________________________________________

On 4 Feb 2010, at 08:05, jspecs wrote:

> Hello,
> I am doing a research project on how pumice rocks change color with  
> varying
> amounts of heat. My findings thus far are that as rocks are exposed to
> higher temperatures, they oxidize and turn darker and darker shades  
> of red.
>
> I am attempting to quantify my data using Image J. I want to be  
> able to
> upload my digital images of the rocks into the program, convert my  
> images
> into RGB images and then determine the mean occurrence of red,  
> green, and
> blue in each image (potentially with color histograms?). My  
> ultimate goal is
> to quantitatively represent the color change occurring as the rocks  
> change
> color with heat by determining the amount of red, green, and blue  
> in each
> image.
>
> Is this possible with this program? Any help would be much  
> appreciated.

[jspecs]

Michael,

<quote author="Michael Schmid-3">

just read the documentation to get a few ideas what could be done:

Image>Color>Split Channels
Analyze>Histogram for each of the channels

or

Image>Type>HSB stack
Analyze>Histogram for each of the stack slices.
Especially the hue slice will be of interest to detect a change of  
hue (see http://en.wikipedia.org/wiki/HSV_color_space if you are not  
familiar with HSB)


I have tried what you mentioned, but both techniques seem to generate slices/channels with varying shades of black, gray, or white. This makes me think that the histogram will just generate its results based on brightness as opposed to RGB. Am I wrong?

[Colin Rickman-3]

Hi

The displayed colour of the individual R, G and B slices is not
important. The shades of grey you see in the image contain the intensity
value for the respective channel (R, G or B). If you recombined them you
would end up back with the original image.

Colin

Dr Colin Rickman
Department of Chemistry (WP 2.03)
School of Engineering and Physical Sciences
Heriot-Watt University
Edinburgh
EH14 4AS

Tel: [phone] (Office)
Tel: [phone]
Fax: [phone]


On 04/02/2010 15:45, jspecs wrote:

> Michael,
>
>
>
> just read the documentation to get a few ideas what could be done:
>
> Image>Color>Split Channels
> Analyze>Histogram for each of the channels
>
> or
>
> Image>Type>HSB stack
> Analyze>Histogram for each of the stack slices.
> Especially the hue slice will be of interest to detect a change of
> hue (see http://en.wikipedia.org/wiki/HSV_color_space if you are not
> familiar with HSB)
>
>
> I have tried what you mentioned, but both techniques seem to generate
> slices/channels with varying shades of black, gray, or white. This makes me
> think that the histogram will just generate its results based on brightness
> as opposed to RGB. Am I wrong?
>
>
>
>    

[Michael Schmid]

Hi,

of course, the histogram of the green channel just contains the  
brightness of this channel. But if you compare, e.g., the histogram  
of the green channel with that of the red channel, a more reddish  
tone will give you am peak at higher values in the red channel and a  
peak at lower values in the green one.

Similarily, the peak in the hue channel of a HSB stack will be  
shifted towards red (i.e. towards 0 if the object is more greenish or  
towards 255 if it is more bluish; in any case, more reddish means  
further away from 128).

Michael
________________________________________________________________

On 4 Feb 2010, at 16:45, jspecs wrote:

> Michael,
>
>
>
>> just read the documentation to get a few ideas what could be done:
>>
>> Image>Color>Split Channels
>> Analyze>Histogram for each of the channels
>>
>> or
>>
>> Image>Type>HSB stack
>> Analyze>Histogram for each of the stack slices.
>> Especially the hue slice will be of interest to detect a change of
>> hue (see http://en.wikipedia.org/wiki/HSV_color_space if you are not
>> familiar with HSB)
>
>
>
> I have tried what you mentioned, but both techniques seem to generate
> slices/channels with varying shades of black, gray, or white. This  
> makes me
> think that the histogram will just generate its results based on  
> brightness
> as opposed to RGB. Am I wrong?
>
>
>
> --
> View this message in context: http://n2.nabble.com/RGB-Imaging- 
> tp4511970p4514230.html
> Sent from the ImageJ mailing list archive at Nabble.com.

[Fabrice Senger]

Colin Rickman a écrit :

> Hi
>
> The displayed colour of the individual R, G and B slices is not
> important. The shades of grey you see in the image contain the
> intensity value for the respective channel (R, G or B). If you
> recombined them you would end up back with the original image.
>
> Colin
>
> Dr Colin Rickman
> Department of Chemistry (WP 2.03)
> School of Engineering and Physical Sciences
> Heriot-Watt University
> Edinburgh
> EH14 4AS
>
> Tel: [phone] (Office)
> Tel: [phone]
> Fax: [phone]
>
>
> On 04/02/2010 15:45, jspecs wrote:
>> Michael,
>>
>>
>>
>> just read the documentation to get a few ideas what could be done:
>>
>> Image>Color>Split Channels
>> Analyze>Histogram for each of the channels
>>
>> or
>>
>> Image>Type>HSB stack
>> Analyze>Histogram for each of the stack slices.
>> Especially the hue slice will be of interest to detect a change of
>> hue (see http://en.wikipedia.org/wiki/HSV_color_space if you are not
>> familiar with HSB)
>>
>>
>> I have tried what you mentioned, but both techniques seem to generate
>> slices/channels with varying shades of black, gray, or white. This
>> makes me
>> think that the histogram will just generate its results based on
>> brightness
>> as opposed to RGB. Am I wrong?
>>
>>
>>
>>    
>

Hi,

I would aswell give a try to the color space converter plugin an not
only work with HSB, maybe Lab would be of interest...
And color spaces are just different ways to depict the data you collected.

Fabrice.