Computer scientists and electrical engineers survey methods for characterizing and evaluating the performance of algorithms used to design computer vision systems. They consider general issues, methodological aspects, statistical aspects, comparative studies, selected methods and algorithms, and domain-specific evaluation in medical imaging. Among the topics are experiences with the empirical evaluation of computer vision algorithms, motion extraction, propagating covariance, evaluating numerical solution schemes for differential equations, unsupervised learning for robust texture segmentation, and error metrics for quantitatively evaluating medical image segmentation. Annotation c. Book News, Inc., Portland, OR (booknews.com)
This book addresses a subject which has been discussed intensively in the computer vision community for several years. Performance characterization and evaluation of computer vision algorithms are of key importance, particularly with respect to the configuration of reliable and robust computer vision systems as well as the dissemination of reconfigurable systems in novel application domains. The objective of this volume is to provide a scientific foundation for performance characterization of computer vision methods, to give an overview of methodologies of comparative assessment of algorithms and to present evaluation approaches for a variety of computer vision applications. This volume comprises six parts: general issues; methodological aspects; statistical aspects; comparative studies; selected methods and algorithms; and finally a domain-specific part on evaluation in medical imaging. Audience: This book can be read by both specialists and graduate students in computer science and electrical engineering who take an interest in computer vision, image processing, and algorithms.
