Exploiting Depth Discontinuities for Vision-Based Fingerspelling Recognition | IEEE Conference Publication | IEEE Xplore