There is one question we get a lot from industrial customers. What is the proper wavelength range to sort my samples? There are two ways to find this out.

First, you could look at literature, scientific articles, or books. But the second, more straightforward way is to make test measurements. I’m going to demonstrate this using two different kinds of samples.

First, fabrics from different colors. There are polyester, cotton, and wool.
The second sample type is nuts, and I will demonstrate how to separate the nut meat from the shell.

So, let’s start.

To demonstrate if an application can be done with an FX17, wavelength range from  900 to 1700 nanometers. I have measured some samples using our laboratory scanner. Known examples include wool, polyester, and cotton. After that, I imported them to a software. In this case, I’m demonstrating using the Perception studio software from Perception Park company, an Austrian Company. This is quite intuitive software, and it gives good insight into what the spectra look like.

First, I select the spectra for each material, first polyester, which I know is here. Then, the second spectra, cotton. Which I know is here and last wool. And we can immediately start looking at the spectra and the possible differences. Indeed, we have pretty distinctive differences in this region and this region, meaning that this application probably or actually can be done using FX17.

We create the application, and then we look at the results. So yellow was wool, blue cotton, and green polyester. On the one side of the image,  we had three fabrics made out of cotton, and they are classified as cotton as the one that I pointed out, although they had different colors.

My second example is pistachio nuts. We want to sort the nut meat from the shell. I have already imported the image and selected the shell areas and nut meat. And when we look at the spectra, we see a lot of intensity changes, and the spectral differences are not that clear.

In this case, I prefer normalization and the first derivative to see the spectral changes. And again, we see that at least in 1200 nanometers and even into longer wavelengths, we have a clear separation of spectra. And if we make the classification, we see that the result seems somewhat reasonable.

We have been able to separate the nut meat from the shells.