Customize last from multiple foot images by a little interaction

To customize last from multiple foot images, this paper presents a pragmatic approach. Not using any laser device, eight images are shot to construct a visual hull, and later the visual hull is refined by some surface features which are extracted from SIFT features. Then the digitized foot is cut into three parts and seven sections. These parts and sections are edited with interaction and finally deformed into a smooth and convex last which adhere to the digitized foot. The method has these novelties: (1) last is constructed from customizer’s foot images, which is fitting to the foot; (2) a unique imaging rack, which is beneficial to taking all-around images and estimating the preliminary model; (3) the digitized foot is cut into three parts and various constrains are applied so that the model is not firm but formable.