Empirical evaluation methods in computer vision / editors, Henrik I. Christensen, P. Jonathon Phillips.

Format:

Book

Contributor:

Christensen, H. I. (Henrik I.), 1962-

Phillips, P. Jonathon.

Series:

Series in machine perception and artificial intelligence ; v. 50.

Series in machine perception and artificial intelligence ; v. 50

Language:

English

Subjects (All):

Computer vision--Evaluation--Congresses.

Computer vision.

Physical Description:

1 online resource (172 p.)

Edition:

1st ed.

Place of Publication:

River Edge, N.J. : World Scientific, c2002.

Language Note:

English

Summary:

This book provides comprehensive coverage of methods for the empirical evaluation of computer vision techniques. The practical use of computer vision requires empirical evaluation to ensure that the overall system has a guaranteed performance. The book contains articles that cover the design of experiments for evaluation, range image segmentation, the evaluation of face recognition and diffusion methods, image matching using correlation methods, and the performance of medical image processing algorithms. <i>Sample Chapter(s)</i><br>Foreword (228 KB)<br>Chapter 1: Introduction (505 KB)<br> <

Contents:

Contents ; Foreword ; Chapter 1 Automated Performance Evaluation of Range Image Segmentation Algorithms ; 1.1. Introduction ; 1.2. Scoring the Segmented Regions ; 1.3. Segmentation Performance Curves ; 1.4. Training of Algorithm Parameters ; 1.5. Train-and-Test Performance Evaluation

1.6. Training Stage 1.7. Testing Stage ; 1.8. Summary and Discussion ; References ; Chapter 2 Training/Test Data Partitioning for Empirical Performance Evaluation ; 2.1. Introduction ; 2.2. Formal Problem Definition ; 2.2.1. Distance Function ; 2.2.2. Computational Complexity

2.3. Genetic Search Algorithm 2.4. A Testbed ; 2.5. Experimental Results ; 2.6. Conclusions ; References ; Chapter 3 Analyzing PCA-based Face Recognition Algorithms: Eigenvector Selection and Distance Measures ; 3.1. Introduction ; 3.2. The FERET Database ; 3.3. Distance Measures

3.3.1. Adding Distance Measures 3.3.2. Distance Measure Aggregation ; 3.3.3. Correlating Distance Metrics ; 3.3.4. When Is a Difference Significant ; 3.4. Selecting Eigenvectors ; 3.4.1. Removing the Last Eigenvectors ; 3.4.2. Removing the First Eigenvector

3.4.3. Eigenvalue Ordered by Like-Image Difference 3.4.4. Variation Associated with Different Test/Training Sets ; 3.5. Conclusion ; References ; Chapter 4 Design of a Visual System for Detecting Natural Events by the Use of an Independent Visual Estimate: A Human Fall Detector

4.1. Introduction

Notes:

All but two contributions are revised papers from a workshop held in 2000.

Includes bibliographical references.

ISBN:

9789812777423

9812777423

OCLC:

879023551