Quantify flower color, redness, yellowness,

imagej

#1

Hi, I am trying to quantify several excised flower petals, hoping some index would measure the redness, or yellowness. My preliminary understanding is that I can split the image in different channels, count the R, G, B under each channel. The intensity of the pixel from each channel would reflect the Redness, Greeness, or Yellowness of samples, this does not imply that they are the numerical reflection of human eyes see. is that correct? I also see people using this expression [(R + B)/2] - G as an approximation of redness. Can anyone help me to clarify this? https://www.dropbox.com/s/0vur5mleph7mkum/S2.tiff?dl=0


#2

Good day,

did you think about the influence of the spectral characteristic of the light source and of the color filters of the camera target?

In general dealing with color ins’t easy!

Regards

Herbie


#3

Hi Herbie

I have not thought about the above you mentioned. All of the samples are taken under the exactly same condition (again in a typical copy scanner, that the samples are usually pressed and covered up, so the external light source influence would be minimal).
If the [(R + B)/2] - G as an approximation of redness, tough I still did not quite get the rational of using it. There must be similar approach to characterize yellow in a similar fashion.

I really appreciate your input.

Best,

Baoqing


#4

I didn’t write of an external light source.
The scanner has a light source and color filtered sensors.

I don’t understand how you define color.
If you define them physically, you will run into several problems when measuring with your setup.

Your formula is not really founded as long as you don’t define what you mean by red.

If you really use a scanner, how comes that you get these shadows?

Good luck

Herbie


#5

I see what you mean now. Even taking the scanner source into consideration, is it reasonable to assume that light source is constant? If it is not, I was hoping that a standard color checker would help correct it, so that the comparisons can be made in a somewhat leveled field. I was thinking if any index summed from the RGB values can be used to reflect the color that human or animal eyes see. Of course, I do not have to stick to this, as long as there is an objective way of measuring it, I would like to hear more from you on this too. I read multiple posts you replied regarding the intensity, I think they may be related to what I am trying to do. But I think I need to read more.

I never notice the shadow, most of the scanner cover is not flat, made of soft material, might not press the specimen evenly? Otherwise, I cannot think of another explanation for this.

Thanks for the input, Herbie.
Baoqing


#6

The following link might also be interesting to measure color and patterns (e.g., with a camera in the field) and what to consider.

An ImageJ plugin is available, too:

http://www.jolyon.co.uk/myresearch/image-analysis/image-analysis-tools/


#7

Mr. Bao
There is a macro named “RGB Profiles tool” available on the macros page accessable thru plugin>macros heading. With it you should be able to figure out what you need by using any of the line tools, since yellow is simply a combination of red and green.
The following is from your example with the blue channel removed. If you use the RGB tool using a long narrow rectangle you will get a readout of just red and green with list of values to perform any math you wish.
Hope this helps, If you need any more help just ask again.
Bob


#8

Thanks, Bob. My question contains two parts:
1, does the [(R + B)/2] - G has a merit to quantify red?
2, If the yellow is simply a function of Red and Green, can R value + G value under these channels be a measure of yellowness?

Thanks!


#9

Bob,

sorry to say, but this is much too simplistic.

Color-analysis is much more complicated and if done correctly, requires defined light sources and a suitable spectrometer!

RGB-data only catches a restricted part of what makes object colors.

Regards

Herbie


#10

Take a look at e.g. this sort of equipment before you start your research;


#11

Herbie,

It isn’t as simple as you seem to think.​

As you should know, the only information you will EVER get from a RGB camera is RGB data and all the information he needs is output from the profiler. Anything he needs to do can be deduced via Mathematics.​

Bob


#12

Mr. Bao,
I’m sorry, but no. Because some sections of your images are bound to include Blue.

By deducing the data via Math. Use the RGB profiler and you will see what I mean. If you need assistance with the math, I will certainly be here.
Bob


#13

Good day Bob!

It isn’t as simple as you seem to think.​

Interestingly, you seem to know what I think .

EVER

Please don’t shout at me!

the only information you will EVER get from a RGB camera is RGB data

Yes, and this is exactly part of the problem.
Others appear to see it similarly.

Anything he needs to do can be deduced via Mathematics.​

I always vote for mathematics but in his case the problem is more on the measuring side.

Regards

Herbie


#14

Hello Bao,

If this is what you were wanting then just reply and I’ll send instructions.

Good Luck
Bob


#15

Hi, Bob

Sorry the picture that Herbie posted is not the one I was trying to quantify. The picture is the one that I linked to my dropbox where there are seven flower petals need to be quantified with the redness and yellowness. Please let me know if you can open the link.

Thanks and really appreciate your help!

Baoqing


#16

Bao,
Apparently I keep getting the wrong image. I keep getting a color chart, is this what you wanted instead of flowers?
I will try to get the image of flowers.
Bob


#17

Hi, Bob

Sorry for the confusing. The dropbox image you opened is exactly the image i want to analyze, the color checker is on the left, on the right side of the image, there are seven excised flower petals. I put it in the dropbox because it is the original image, it is about over 100M, so it would be hard to send to the mailbox. Please let me know if you can find the samples on the right side of the tiff image. If not, I will compress it and send it through email.

Best,

Baoqing