A new approach for fabric hand prediction with a nearest neighbor algorithm-based feature selection scheme

Fabric hand refers to the subjective perception of fabric mechanical properties. It is an important factor in fabric selection. In the literature, various instruments and tools have been developed to evaluate fabric hand by means of the related physical and mechanical properties. These works have studied the prediction of fabric specimens based on the fabric hand descriptors via either traditional statistical methods or artificial intelligence methods. The artificial neural network (ANN) model and its variation, the fuzzy neural network (FNN)-based intelligent fabric hand prediction system have been studied previously, and they show good accuracy in the forecasting task. The FNN model is especially useful as it provides good prediction accuracy and understandable rule-set at the same time. However, real-world application of the FNN model is restricted because of the great burden of required computation. In this paper, a nearest neighbor algorithm is used for feature selection to reduce computation burden. An improved FNN-based fabric hand prediction model is devised. Data sets collected from 30 participants’ evaluation on a set of 20 fabric specimens, taken from the data source used by Lau et al., are used to verify the proposed system; the system is found to be capable of carrying out fabric hand predictions with many input parameters quickly and with good accuracy.