| Foreword | p. v |
| Contents | p. vii |
| Automated Performance Evaluation of Range Image Segmentation Algorithms | p. 1 |
| Introduction | p. 2 |
| Scoring the Segmented Regions | p. 2 |
| Segmentation Performance Curves | p. 4 |
| Training of Algorithm Parameters | p. 6 |
| Train-and-Test Performance Evaluation | p. 9 |
| Training Stage | p. 12 |
| Testing Stage | p. 15 |
| Summary and Discussion | p. 18 |
| References | p. 21 |
| Training/Test Data Partitioning for Empirical Performance Evaluation | p. 23 |
| Introduction | p. 23 |
| Formal Problem Definition | p. 25 |
| Distance Function | p. 26 |
| Computational Complexity | p. 27 |
| Genetic Search Algorithm | p. 28 |
| A Testbed | p. 30 |
| Experimental Results | p. 32 |
| Conclusions | p. 35 |
| References | p. 36 |
| Analyzing PCA-based Face Recognition Algorithms: Eigenvector Selection and Distance Measures | p. 39 |
| Introduction | p. 40 |
| The FERET Database | p. 41 |
| Distance Measures | p. 41 |
| Adding Distance Measures | p. 42 |
| Distance Measure Aggregation | p. 43 |
| Correlating Distance Metrics | p. 44 |
| When Is a Difference Significant | p. 45 |
| Selecting Eigenvectors | p. 49 |
| Removing the Last Eigenvectors | p. 50 |
| Removing the First Eigenvector | p. 51 |
| Eigenvalue Ordered by Like-Image Difference | p. 52 |
| Variation Associated with Different Test/Training Sets | p. 53 |
| Conclusion | p. 56 |
| References | p. 59 |
| Design of a Visual System for Detecting Natural Events by the Use of an Independent Visual Estimate: A Human Fall Detector | p. 61 |
| Introduction | p. 61 |
| Approach | p. 62 |
| Data Collection | p. 65 |
| Velocity Estimation | p. 65 |
| Colour Segmentation and Velocity Estimation | p. 66 |
| IR Velocity Estimation | p. 70 |
| Velocity Correlation | p. 72 |
| Data combination | p. 73 |
| Conclusions | p. 79 |
| Neural Network Fall Detector | p. 80 |
| Data Preparation, Network Design, and Training | p. 80 |
| Testing | p. 81 |
| Conclusions | p. 85 |
| References | p. 87 |
| Task-Based Evaluation of Image Filtering within a Class of Geometry-Driven-Diffusion Algorithms | p. 89 |
| Introduction | p. 89 |
| Nonlinear Geometry-Driven Diffusion Methods of Image Filtering | p. 90 |
| Diffusion-Like Ideal Filtering of a Noise Corrupted Piecewise Constant Image Phantom | p. 93 |
| Stochastic Model of the Piecewise Constant Image Phantom Corrupted by Gaussian Noise | p. 95 |
| Estimates of Probability Distribution Parameters for Characterization of Filtering Results | p. 96 |
| Implementation results | p. 101 |
| Conclusions | p. 108 |
| References | p. 113 |
| A Comparative Analysis of Cross-Correlation Matching Algorithms Using a Pyramidal Resolution Approach | p. 117 |
| Introduction | p. 118 |
| Area Based Matching Algorithms | p. 119 |
| Cross-Correlation Algorithms | p. 122 |
| Pyramidal Processing Scheme | p. 125 |
| Number of Layers | p. 127 |
| Decimation Function | p. 127 |
| Matching Process | p. 128 |
| Interpolation | p. 129 |
| Disparity Maps | p. 130 |
| Experimental Results | p. 130 |
| Experiment Layout | p. 130 |
| Disparity Maps | p. 132 |
| Disparity Error | p. 133 |
| Computational Load | p. 136 |
| Conclusion | p. 141 |
| References | p. 141 |
| Performance Evaluation of Medical Image Processing Algorithms | p. 143 |
| Introduction | p. 143 |
| Presentations | p. 144 |
| New NCI Initiatives in Computer-Aided Diagnosis | p. 144 |
| Performance Characterization of Image and Video Analysis Systems at Siemens Corporate Research | p. 145 |
| Validating Registration Algorithms: A Case Study | p. 146 |
| Performance Evaluation of Image Processing Algorithms in Medicine: A Clinical Perspective | p. 150 |
| Performance Evaluation: Points for Discussion | p. 152 |
| Panel Discussion | p. 154 |
| References | p. 158 |
| Table of Contents provided by Ingram. All Rights Reserved. |
