An application of computer vision to lace cutting

Abstract

Cutting cloth and sewing garments are largely manual operations in many clothing companies. Difficulties of automating the manipulation of limp material which tends to stretch, shrink and wrinkle have delayed the introduction of automation into the apparel industry. We describe the application of a low-cost computer vision system to the automatic cutting of lace. Lace must be cut into pieces before sewing onto a garment. The point where the lace pattern is cut must be consistent so that the garments have a similar appearance. Because lace tends to stretch it is not possible to use dead reckoning to cut the lace at a consistent position in the lace pattern. The function of the vision system is to detect a reference feature in the lace pattern so that variations in the lace position can be accommodated. We present details of a prototype system for this application and some results of applying it to lace samples.