Customer Reviews

Pattern Classification and Scene Analysis

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

 

 

This book was published in 1973 and there have been many advances since. Still I find it provides great exposition of the fundamental concepts. In fact the nearest neighbor algorithms that are now popular are covered in this book and date back to the work of Cover and Hart in the late 1960s. Those new to pattern recognition who think kth nearest neighbor rules are new should read this book to find out exactly when it was really thought up.
For a more up-to-date treatment, see McLachlan's recent book in the Wiley statistics series. However, this book provides valuable explanations of Bayes rules and shows pictorially what the boundaries look like for linear and quadratic classifiers. In fact I borrowed their pictures in Chapter 2 of my book on bootstrap methods and it appears in both the first and second editions of my book.

The authors with the help of a third author have updated this book recently and I highly recommend the new addition which maintains many of the nice features of the original.

While it's becoming a little dated (no problem, v.2 is in the works), this is the "must have" book for anyone concerned with pattern classification and scene analysis - not just computer vision, but anything related to pattern classification. It introduces all of the basic techniques and approaches that anyone involved in data reduction and classification should be familiar with: contour search, interval splines, chain encoding, the effects of digitizing the input data, pattern representation and classification, etc. Get it. You'll be glad you did. -Drew