Customer Reviews

Vision: A Computational Investigation into the Human Representation and Processing of Visual Information

5 star:		(3)
4 star:		(2)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

Although dead at the young age of 36 from leukemia, Marr's computational and mathematical approach to vision revolutionized the entire area of vision research, after which it was never the same. There are strong hints of this approach in the earlier work of Julesz and Gibson, but Marr's work takes the whole field a quantum leap further, giving it a rigorousness and mathematical elegance never before seen.

For example, to mention just a few of his important ideas, Marr's demonstrations that retinal receptive field geometry could be derived by Fourier transformation of spatial frequency sensitivity data, that edges and contours could be detected by finding zero crossings in the light gradient by taking the Laplacian or second directional derivative, that excitatory and inhibitory receptive fields could be constructed from "DOG" functions (the difference of two Gaussians), and that the visual system used a two-dimensional convolution integral with a Gaussian prefilter as an operator for bandwidth optimation on the retinal light distribution, were more powerful than anything that had been seen up to that time.

It was as if vision research suddenly acquired its own Principia Mathematica, or perhaps General Relativity Theory, in terms of the new explanatory power Marr's theories provided. Truly an extraordinary book from an extraordinary thinker in the area of perception, vision, and the brain.

This book has quickly become a classic since its publication in 1982. It offers an innovative theoretical approach to explain what it means when we say that we "see" something. Due to Marr's particular interests, his approach also offers, to those interested in creating machine vision that mimics human vision, potential algorithms for doing so. His breadth and depth of knowledge in mathematics, psychology, neurophysiology, and engineering allowed him to integrate the fields in a way never done before. His untimely death was a tragic loss for us all. Although this book may be a bit difficult to understand for the typical psychology student studying human visual perception, it really belongs on his/her shelf. I agree with Marr in his statement that ignorance of the necessary mathematics is inexcusable. For the typical mathematics or engineering student, the book will probably read fairly easily. Although I represent the typical psychology graduate student, it's apparent from my exchanges with my computer science friends working in artificial intelligence that the theories in this book are very well accepted in their academic circles as well.

It is quite true: this is probably the best book on vision ever published. David Marr combined an incredible depth and width of knowledge in all neccessary fields (psychology, biology, mathematics, computer science) into this book. You will need some background in mathematics before you can fully appreciate Marr's theories. (Notably Fourier transforms) The most important thing about this book is that it represents ONE paradigm to solve problems of perception, and proves that this paradigm works in a number of cases. Unfortunately, David Marr did not live long enough to implement his paradigm in all aspects of vision. Had he still been alive, we would be looking at a much more unified field of vision research today.

I have been shocked three times after reading chapter one only of this book three years ago when I firstly contacted genius David Marr in my mind. It was because of i) his comprehensive understanding about human visual perception, ii) he was undoubtly young on the contrary to his comprehensiveness, and iii) regretfully he had gone young at his vital age. Until now, it is hard for me to deny his influences in directing my studies in vision. Don't say about vision before reading this book ! This is the best book in vision among those published until now at least in my sight.

This book is interesting in the respect that it takes a biological view on computational vision. I do recommend it for that perspective which is certainly valuable. It does not have fair coverage of biologically inspired vision, nor does it correctly show the author's work in relation to others.