Who first described vision as "inverse graphics"? It is often ascribed to Horn in either his 1977 paper on Understanding Image Intensities or his 1986 book, Robot Vision. It is not in the latter and in the former Horn writes of his physics-based approach "I do not mean to suggest analysis by synthesis. Nothing of the sort!" So it wasn't Horn. Poggio certainly advanced this view and wrote early about inverse problems in general; he used the term "inverse optics" in 1988 but focused on regularization rather than graphics. Grenander (in the '70's) is clearly the source for "analysis by synthesis", but I want the first published reference to term "inverse graphics". I don't think it was Marr. Vision historians: do you know?
The first reference is actually earlier; Baumgart's PhD Thesis is about an "inverse computer graphics approach to computer vision". Geometric Modeling for Computer Vision, Stanford AI Lab Memo AIM-249, Oct 1974.
Add a comment...