LabVIEW
Overlaying intensity graphs to create a picture
AntLee29 a écrit:
I have a 16x256x256 3-dimensional array of intensity images which represent 16 different 256x256 intensity graphs. Each page of the stack represents a single wavelength on the visible spectrum. Is it possible to create a full 16-color picture by overlaying the intensity graphs together? If so, how would I go about doing that? I have attached the 3-d stack of data.
Since a color image is made with only 3 colors (RGB), you need a way to re-code your data. The 16 wavelengths (WL) could be considered as belonging to the visible spectrum, between 400 and 800 nm, and a simple way to convert the WL as RGB would be to use Dan Burton's linear approximation, as described here. Alternatively, you could define you own arbitrary colors...
Attached is an example result using D Burton color coding.
Message Edité par chilly charly le 08-02-2008 02:48 PM
*He's the Graphing Wizard, there has to be trick,
He's the Graphing Wizard, he's got such a supple wrist
How do you think he does it? I don't know
What maeks him so good?
....
I pass my graphing crown to him, to him!
Ben
* To be read with the rock-opera "Tommy" by the Who playing in the background.
Ahh you caught me in the middle of editing my post. 
Correct me if I'm wrong, but it looks like the equalizer lets you bias the color table on each wavelength to make it stand out more. That means I should keep them all at the same level to get the normal distribution?
I've got the rest of the vis understood now.
I suppose I'll need to manually figure out an RGB value for wavelengths under 380, as the first 2 pages in the stack correspond to 359nm and 378nm.
You're looking at endothelial cells in a fixed collagen matrix. We used a laser to image the sample and a prism to spread the fluorescence across a 16-channel photomultiplier tube, which was tuned to the wavelength range that we desired.
Message Edited by AntLee29 on 08-03-2008 03:01 AM
Correct me if I'm wrong, but it looks like the equalizer lets you bias the color table on each wavelength to make it stand out more. That means I should keep them all at the same level to get the normal distribution?
I've got the rest of the vis understood now.
I suppose I'll need to manually figure out an RGB value for wavelengths under 380, as the first 2 pages in the stack correspond to 359nm and 378nm.You're looking at endothelial cells in a fixed collagen matrix. We used a laser to image the sample and a prism to spread the fluorescence across a 16-channel photomultiplier tube, which was tuned to the wavelength range that we desired.
Message Edited by AntLee29 on 08-03-2008 03:01 AM
AntLee29 a écrit:
(...) Correct me if I'm wrong, but it looks like the equalizer lets you bias
the color table on each wavelength to make it stand out more. That means I should keep them all at the same level to get the normal distribution?
I've got the rest of the vis understood now.
You're looking at endothelial cells in a fixed collagen matrix. We used a laser to image the sample and a prism to spread the fluorescence across a 16-channel photomultiplier tube, which was tuned to the wavelength range that we desired.
I don't think there is something such as an "normal" distribution of light intensities in your application : if all the sliders were set at unity, you would get a color spectrum like the one below for intensities values all fixed at 255. Now, you have to consider that your signal value range is not absolute, specially because there is no color equivalence between UV and visible. Accordingly, the equalizer was used to optimize the look-and-feel of the resulting image. Forget about "normal distribution" (anyway you are not making measurements on the displayed picture !) and focus only on the image "quality".
The wavelength_to_RGB vi has been improved by introducing a fall-off correction at both ends of the spectrum. This makes the vi closer to Dan Burton's C program. I have also added a "gamma" exponent, to account for the fact that the eye/display system is not linear. With gamma < 0, the lower intensities are boosted up, but this also increases the noise. The sliders are now coerced in the range 0.5-10, making the display optimization slightly easier.
The new version is attached.
One question : are the 359 and 378 emission or excitation wavelengths ?
Message Edité par chilly charly le 08-03-2008 02:20 PM
