Machine Perception of Three-Dimensional Solids
Lawrence Roberts showed how to recover the three-dimensional shape of a polyhedron from a single photograph, going from intensities to edges to a model.
Why it matters
Computer vision got its first complete scheme: an image is turned into a description of a scene rather than classified whole.
Roberts found intensity changes with an operator later named after him, joined them into lines, and lines into faces of known solids. The constraints were severe: polyhedral scenes, good lighting, a known set of shapes. Posing the task as a sequence of levels of description shaped the field for decades.