tranparency of a liquid and colour comparison

[ziedhosni27]

Hi Udka,

You can find attached my real samples.I tried to put a white and a blue background.Actually the colour is not as important as the tranparancy of the sample.  Always I record the samples with a HD webcam then I analyse the ROI of each one.

You can notice the real transparance of the samples (for 1 to 7 from the left to the right). I used another analytical method to measure the transparence.

Sample Ref.	%T
1	2
2	-
3	1
4	78
5	51
6	23
7	>90

[ziedhosni27]

Hi,

Please can someone help me to sort this problem out. Do you have an idea to use the webcam and vision kit of Labview to get values similar or close to those of the previous table.

Thanks for your help.

[udka]

Hi ziedhosni,
-Can i know how did you get those values above?
-When i saw the images, under blue background, it is very clear that they are bottles 4 and 7 are transparent, and 5 little also, because i can see blue color behind it. which means we can detect blue color intensity and give proper transparency.
-But for others it is very difficult isn't it ?(this is just my understanding).
-One more thing is the labels on the bottles can give false meaning also. I think we should do the analysis without labels on the bottles.
-Color may not be important, but the transparency can be defined when you see through it isn't it? So, if you have yellow background can we decide the transparency? (Which i think is most difficult).
-So, is it possible to first see yourself under which color background, we can see through properly and then post image?
-And you need to have same background for all bottles, may be i prefer instead of taking all the bottles at a time, keep 2-3 bottles so that background is uniform and covered through out.
-When you kept blue background, lighting is different when image was taken with white background.Lighting is trivial task here.and you should have constant lighting.

-Lets see if anyone else has better ideas.

[ziedhosni27]

Thank you again Udka,

Thoese value were measured by someone else using Uv light absorption using the formula:I/I0 where I is the intensity of light.

For the blue I think it is the best condidate as it is the complementary for the orange formed from red and yellow.

 So do you know how to measure the intensity of blue ( I mean which unit and is it possible to make a correlation between this measurement and the tranparency values of the table and then make the interpolation. I explain. If the sample give 200 blue and the trasparancy  is 60% and another sample the blue is 0 and the transparancy is 0 ,then an unknown sample with a blue equal to 100 means that the transparancy is 30%. Do you think this is do-able.

Here I put all the samples together, but in reality the measurement is done one by one. The big problem I think is the intensity of light in the room that affect too much the colours detected by the camera.

Have you an idea how can I handle this problem
.

Thanks for your help.

[udka]

Thanks for the reply again.

-Getting blue color values is easy. i am attaching example which i tried for getting the blue values.

-Each point will have Red(0-255),Blue(0-255) and Green(0-255). See in the example i added.

-The green rectangle is the selected ROI(Region of interest), where you have blue value of around 200. means if you take proper blue(try it in paint) and when you check it will be 255(Max).

-So, if i get the same value at bottle also, then it is completely 100% transparent(am i right?)

-But, when i keep same ROI at bottle, i get value around 20-30 (which is mix of yellow also). or because of disturbance it is not clear??

-Yeah interpolation of values also possible but before that we need to check which values we should take.

-Is it possible to take only one bottle(high transparent) with blue background and then see values?

-I am attaching vi also, please try to use and see if it is helpful.

-These kind of experiments are highly lighting related. So you should have controlled lighting which will be constant.Are you using any external light or normal ambient light?

-Sorry i am not expert in these, but i will try if i can help.

[ziedhosni27]

You are really nice and helpfull.

Thanks for the exampl, it clrify lot of things.

By default, I think colours can just divided into blue green and red. But that makes me confused if I select the yellow for example, I found it is 0 blue 250 red and 250green. How this can be?

-Is it possible to take only one bottle(high transparent) with blue background and then see values?

For this question, I can use the background alone as a reference, I think even if there is a noise in the lighting, the comparison with the reference will correct the value.

Also, I am thinking to add like an intense flash light to the camera to be sure I am working in the same conditions of light like in the professional cameras using flash in every snapshot.

Best,

[udka]

--Actually it will be R-255, G-255, and B-0 for yellow (Ideal).
-If you mix two colors obviously you will get different color. And if you combine all of them you get white(R-255,G-255,B-255).
-Please check this link for color values of different colors.
http://www.rapidtables.com/web/color/RGB_Color.htm
-Be careful with flash light, if the light is too much, then also colors might change.