Classification of Hand Based on Fabric Drape

Abstract

In this paper fabric hand was objectively evaluated through the exploration of three-dimensional drape. Four Windows Kinect sensors were arranged to capture the point cloud data of the draped fabric. To extract the actual drape features, a new algorithm was designed and implemented in Matlab technical computing environment. Thirty seven woven fabrics were clustered in accordance with subjective hand preferences of softness and stiffness collectively as ground truth. The performance of the objectively clustered data was evaluated against the ground truth. The results revealed that the objective classification of the fabrics using 3D drape data correlated to the fabric hand preferences with 86 % matching accuracy when applied on same fabrics. The evaluation of hand from fabric drape in a non-touch manner will be a better reference for online shopping and computer simulation.