Geometry is basic to computer vision and contains many subareas. We are well aware that this page does not yet contain references on sufficient areas.
The best single reference is a paperback by Roe. It is very clear and has the most appropriate topic coverage. The book by Jennings is attractive and provides a wider perspective on the relations to math and science. The book by Smart is also clear, more expensive, and has a slightly different range of topics, tending more towards alternative geometries.
The classic book by Hilbert and Cohn-Vossen is still in print and very readable. Great for getting ideas, but useless for quantitative details.
A clear presentation of basic facts from spherical geometry can be found in the book by Jennings (see above) and in McCleary's differential geometry text.
The book by Hahn is absolutely wonderful and assumes almost no background. A useful preliminary if you are having trouble reading more advanced material using complex techniques.
The standard reference for tilings and patterns in 2D is the fat volume by Grünbaum and Shephard. It has everything, in detail, with lots of pictures, and it is extremely readable. Be careful: there is also an abridged edition, which is significantly less nice.
The Hargittai book is an inexpensive paperback with many wonderful photos of objects illustrating different types of symmetries. You've seen similar photos in other places, but not so many and such nice photography.
A particularly clear, readable, comprehensive, and cheap introduction to differential geometry can be found in:
Computational geometry is sufficiently central to computer vision that it deserves its own page.