Page 28 - Read Online
P. 28
Page 112 Ding et al. Art Int Surg 2024;4:109-38 https://dx.doi.org/10.20517/ais.2024.16
Figure 1. The enabling role of geometric scene understanding for digital twin. All figures in the article are generated solely from the
authors’ own resources, without any external references.
Figure 2. An overview of geometric representations. PCA: Principal component analysis; SDF: signed distance function.
with a trade-off between the error introduced by quantization and the computation and memory
complexity . The high computation and storage complexity hinders scaling up the accuracy and range of
[34]
the representation. More importantly, there is an obvious gap between discrete representations and
humans’ intuition of expressing geometry from a semantic level . Humans perceive geometry in its
[34]
entirety or connection between parts, whereas grid-based discrete representation represents geometry
mostly by the spatial aggregation of the cells without explicit relations among them. Grid-based
representations are suited well for tasks that focus on individual cell values/patterns in a group of cells.
