Math Maps How Faces Transform
At its core, shape is what remains when you discount the effects of location, orientation, and overall size. Imagine a photograph of a skull. If you rotated it, moved it around your screen, or zoomed in, the shape itself hasn't changed – just how it's presented.
3D CT scan showing craniofacial landmarks used in shape analysis
Component | Average % of Total Shape Variance | Key Characteristics |
---|---|---|
Affine | 25-35% | Uniform scaling, stretching, shearing. Parallelism preserved. |
Non-Affine | 65-75% | Localized bending, twisting, remodeling. Represents specific biological growth processes. Dominant component. |
Landmark Region | Approx. Procrustes Distance | Direction of Dominant Change |
---|---|---|
Menton (Chin) | 0.12 | Strong Forward (+Z) & Slight Downward (-Y) |
Gonion (Jaw Angle) | 0.08 | Lateral Outward (+X/-X) & Downward (-Y) |
Nasion (Nasal Root) | 0.05 | Forward (+Z) & Slight Upward (+Y) |
Glabella (Brow) | 0.04 | Minimal Change / Slight Posterior (-Z) |
Basion (Skull Base) | 0.02 (Reference) | Minimally displaced (Used as reference) |
Captures high-resolution spatial data of specimens (CT, MRI, Laser Scanner).
Precisely defined, biologically homologous points (e.g., suture intersections, bone tips).
Allows researchers to pinpoint and record 3D coordinates of landmarks on digital images.
Performs Procrustes superimposition, statistical shape analysis, and visualization.