NIRS identification of black textiles : Improvements for waste textiles sorting
Rintala, Niko (2019)
Julkaisun pysyvä osoite on
Near infrared spectrometry (NIRS) can be harnessed to identify organic compounds by building libraries of reference spectra with validated samples. In Lahti University of Applied Sciences (LAMK), NIRS is utilized in a pilot-sized textile sorting line to identify the composition of samples. Comparing sample spectra to spectral libraries, the sample material may be identified. The libraries currently cover homogeneous wool, polyester, cotton and viscose, with more under construction. NIRS generally ignores colorants, but black textiles have proven to occasionally share spectral anomalies that make their material identification impossible with the current algorithm. The purpose of this thesis was to find improvements for the current pilot-sized textile sorting line of LAMK. The focus was on black textiles and how to potentially improve sorting capabilities of such materials. A test run with a limited supply of black textiles was set up. It seems that with the use of NIRS, it would be possible to improve the detection of black samples in the sorting line. However, further testing with more verifiable samples is required to increase identification accuracy. The result of the test is an identification method that due to its inaccuracy claims small percentile blends to be pure materials. Other means to deal with small percentile blends were studied in the thesis. With high enough resolution, a camera capable of shape recognition could potentially detect the weave of textiles, expanding sorting potential by possibly filtering high risk textile weaves out of the stream.