Fibersort® technology

Fibersort® is a new technology that enables the recycling industry to identify and separate textiles based on fiber composition and color properties.

The technology is based on a combination of NIR (Near-Infrared) and RGB camera technology.

• NIR camera is used to analyze the fiber composition
• RGB camera is used to analyze the color composition

We currently have 6 fiber types trained to the system (wool, cotton, polyester, viscose, acrylic and nylon) but can be extended to other fibers if these are trained to the textiles library.

NIRS detection

The fiber detection technology we use is NIRS, which stands for Near Infra-Red Spectroscopy.
It is a spectroscopic technique based upon molecular absorptions measured in the Near Infrared part of the spectrum. It is sensitive to organic constituents and can provide both chemical and physical information.

NIRS is based on the fact that molecules absorb specific frequencies that are characteristic of their structure. The system measures the absorption of radiation as a function of the wavelength.

Using advanced machine learning algorithms, it can determine the fiber composition in textiles, virtually predicting the label that is or should be attached to the textiles. The large variety of textiles caused traditional software and sorting machines to fail in this application. A huge data set is really important to ensure accurate sorting. Valvan’s team is very committed to keeping this data base up to date together with the users of the Fibersort® machines.

Color sorting

Textile dying is a very polluting activity with a lot of chemicals involved. If the recycling process of the textiles can avoid having to dye again, the process can have a huge impact on the environment. The Fibersort® is equipped with an RGB camera enabling accurate sorting in different colors. The color sorting works on two levels:

1. Single versus multi-color
2. Precise single color sorting

Limitations

As with any technology, this system also has limitations.

1. The Fibersort® algorithms return an estimation of the fiber concentration in a textile. However this result is not always perfect and will depend on the complexity of the textile. Textiles with complicated blends of multiple fiber types (3 or more), or very atypical concentrations are harder to sort accurately.
   
   
2. This technology is a surface scan
   The scanner is not able to scan through the surface of the textile to see what’s underneath it.
   The Fibersort® process focuses on sorting monomaterial textiles.
   
   
3. Since the technology is based on absorption of light, the light colors have better recognition than the dark colors. So black colored textiles might give an inaccurate result.
   
   
4. Wet textiles will not be accurately scanned because liquids interfere with the spectrum analysis. When we detect such a spectrum, we could blow them off in a separate category, so that they can be sent for drying, provided there are enough blower points.
   
   
5. Separation by color only works for single-colored textiles.
   Multi-colored textiles can only be separated from single-colored textiles.