Abstract: In order to identify the materials of blended yarns, the article first selected the indicators that can reflect the morphological characteristics of wool and viscose fibers in the cross-section images of blended yarn by using MIV-BP method. Then it used probabilistic neural network/support vector machine/extreme learning machine to identify wool and viscose materials. The identification results show that by using suitable feed-forward neural network, the materials can be quickly identified with accuracy up to about 90% and the method has the potential for practical use.

Key words: feed-forward neural network; material identification; wool; viscose; MIV; image processing