Broad, but a mess.

This is, from what I can tell, just about the most complete up-to-date text in the field of computer vision as of late 2006.

But it's a mess.

I'm a PhD student, and have worked my way through more than my fair share of high-level computer science textbooks. This one makes me really appreciate many of them. It reads like a first draft -- overly wordy at times, skipping over important issues, poorly organized... Some concepts that ought to be really simple are made very painful due to what seems to be laziness on the part of the editor. It's like the only people that critiqued this book prior to publication already knew all there is to know about computer vision.

A particularly nasty aspect of this book is that the authors have a horrible habit of including some term in some complex mathematical formula, with no reference whatsoever to that term in the surrounding text! In an explanation of how to use Expectation Maximization in line-fitting, they include a standard-deviation term, with no mention of how you're supposed to choose a value for it other than "...for sigma as before". The only "before" in which the SD (sigma) is mentioned in a similar context that I can find is IN THE PREVIOUS CHAPTER!!!

Anyway, if you want to try to teach yourself vision, don't bother. If you need the book for a class, I'm sorry it's so expensive. Either way, don't expect much.

Other Reviews

• This book is very difficult to read, specially mathematical notation. Authors use some strange notations very awkwad expression!
• complete mess
• Terrible book, Hard to understand!
• Broad, but a mess.