Computational theories of vision

This exciting and original book describes the work of David Marr and his colleagues at MIT on the computational theory of vision, particularly the early stages of shape analysis. Although much of the theorising is frankly speculative and not certain to survive without major modification, it sets new standards of rigour in its formal approach and is certain to have a stimulating effect upon the field. Sadly, Marr's death from leukaemia at the age of 35 means that he will not himself be able to take part in the further development of his ideas, but the book itself will exert an important influence for some time to come. The first point to make, for those who may already have attempted Marr's lengthy and difficult research papers, is that the book is clearly and entertainingly written: It can be tackled by the non-specialist who wishes to see why Marr's work has caused such interest in the vision community. Despite the very difficult circumstances in which the book must have been written, it is carefully put together and gives a fascinating insight into the way in which the author's ideas developed.