| Next Generation Multimedia | |
| Drag-and-Drop Pasting | p. 3 |
| Introduction | p. 4 |
| Related work | p. 5 |
| Optimal Boundary | p. 6 |
| Poisson image editing | p. 6 |
| Boundary energy minimization | p. 7 |
| A shortest closed-path algorithm | p. 8 |
| Fractional boundary | p. 10 |
| A blended guidance field | p. 10 |
| Experimental Results | p. 12 |
| Conclusion and Discussion | p. 14 |
| Motion Patches: Building Blocks for Virtual Environments Annotated with Motion Data | p. 17 |
| Introduction | p. 17 |
| Background | p. 18 |
| Data Acquisition and Processing | p. 19 |
| Motion Patch Construction | p. 22 |
| Stitching Motion Patches | p. 24 |
| Tilable Patches | p. 25 |
| Animation and Control | p. 26 |
| Experimental Results | p. 27 |
| Discussion | p. 29 |
| Video Completion by Motion Field Transfer | p. 35 |
| Introduction | p. 35 |
| Related Work | p. 36 |
| Overview of Our Approach | p. 37 |
| Video Completion Algorithm | p. 37 |
| Local motion estimation | p. 38 |
| Dissimilarity measure of motion vectors | p. 38 |
| Motion Field Transfer | p. 38 |
| Color Propagation | p. 39 |
| Experimental Results | p. 40 |
| Results of Video Completion | p. 40 |
| Another Application: Frame Interpolation | p. 43 |
| Discussion | p. 44 |
| Conclusion | p. 45 |
| Correlative Multi-Label Video Annotation | p. 47 |
| Introduction | p. 47 |
| First Paradigm: Individual Concept Annotation | p. 48 |
| Second Paradigm: Context Based Conceptual Fusion Annotation | p. 49 |
| Third Paradigm: Integrated Multi-label Annotation | p. 49 |
| Our Approach-CML | p. 50 |
| A Multi-Label Classification Model | p. 50 |
| Learning the Classifier | p. 52 |
| Connection with Gibbs Random Fields for Multi-Label Representation | p. 53 |
| Implementation Issues | p. 55 |
| Interacting concepts | p. 55 |
| Concept Label Vector Prediction | p. 57 |
| Concept Scoring | p. 58 |
| Experiments | p. 59 |
| Data Set Description | p. 59 |
| Experiment Setup | p. 60 |
| Experiment Results | p. 61 |
| Conclusions and Future Works | p. 62 |
| References | p. 63 |
| Stereoscopic Video Synthesis from a Monocular Video | p. 65 |
| Introduction | p. 65 |
| Related Works | p. 66 |
| Overview | p. 67 |
| The Cost Function | p. 69 |
| Stereo Cost | p. 70 |
| Similarity Cost | p. 72 |
| Continuity Cost | p. 72 |
| Optimization | p. 73 |
| Initialization | p. 74 |
| Speed-up | p. 74 |
| Optimization for Visual Smoothness | p. 76 |
| Results and Discussions | p. 76 |
| Conclusions | p. 78 |
| References | p. 80 |
| ShapePalettes: Interactive Normal Transfer via Sketching | p. 83 |
| Motivation | p. 84 |
| Contributions and Related Work | p. 85 |
| Interacting with Shape Palettes | p. 86 |
| Shape Palettes in Use | p. 88 |
| Discussion and Limitations | p. 88 |
| Summary | p. 89 |
| References | p. 91 |
| Adaptive Directional Lifting-based Wavelet Transform for Image Coding | p. 93 |
| Introduction | p. 94 |
| 2D Wavelet Transform via Adaptive Directional Lifting | p. 96 |
| ADL structure | p. 96 |
| Subpixel Interpolation | p. 99 |
| R-D Optimized Segmentation for ADL | p. 101 |
| Experimental Results and observations | p. 102 |
| Conclusions and future work | p. 107 |
| References | p. 108 |
| Networking and Systems | |
| Low-Power Distributed Event Detection in Wireless Sensor Networks | p. 113 |
| Introduction | p. 113 |
| Related Work | p. 114 |
| Paper Organization | p. 115 |
| Analysis of Detection and Lifetime | p. 115 |
| System Model | p. 115 |
| Detection Analysis | p. 116 |
| Tradeoff Characterization | p. 117 |
| CAS: Coordinated Wakeup Scheduling | p. 118 |
| Overview | p. 118 |
| Distributed Scheduling Coordination | p. 119 |
| Aggressive Wakeup Adjustment | p. 120 |
| Performance Evaluation | p. 120 |
| Experiment Setting | p. 120 |
| Results | p. 120 |
| Conclusion | p. 123 |
| Understanding User Behavior in Large-Scale Video-on-Demand Systems | p. 125 |
| Introduction | p. 125 |
| The PowerInfo VOD System | p. 126 |
| User Access Patterns | p. 128 |
| User Accesses over time | p. 129 |
| User Arrival Rates | p. 130 |
| Session Lengths | p. 132 |
| Implications | p. 135 |
| Popularity and User Interest | p. 135 |
| Pareto Principle | p. 135 |
| Request (Popularity) Distribution | p. 136 |
| Rate of Change in User interest | p. 139 |
| Understanding Popularity | p. 141 |
| Introduction of New Content | p. 141 |
| Impact of Recommendations | p. 142 |
| Related Work | p. 145 |
| Conclusion | p. 146 |
| References | p. 147 |
| A representation theorem for minmax regret policies | p. 149 |
| Introduction | p. 149 |
| Three qualitative decision criteria | p. 150 |
| When does a policy have minmax regret representation? | p. 151 |
| Conclusions | p. 154 |
| References | p. 154 |
| A Note on the Cramer-Damgard Identification Scheme | p. 155 |
| Introduction | p. 155 |
| Description of the Cramer-Damgard Intended-Verifier ID Scheme | p. 156 |
| [Sigma]-protocol and [Sigma subscript OR]-protocol | p. 156 |
| Description of protocol | p. 157 |
| Two Man-in-the-Middle Attacks | p. 157 |
| The replaying attack | p. 157 |
| The interleaving attack | p. 158 |
| Concluding Remarks | p. 158 |
| References | p. 159 |
| Natural User Interface | |
| A Tree-based Kernel Selection Approach to Efficient Gaussian Mixture Model - Universal Background model based Speaker Identification | p. 163 |
| Introduction | p. 163 |
| GMM-UBM identification system | p. 165 |
| Tree-based kernel selection | p. 167 |
| Distortion measure & cluster centroid | p. 167 |
| Tree construction | p. 168 |
| Mixture components selection | p. 168 |
| Tree structure & algorithm efficiency | p. 169 |
| Observation reordering based pruning | p. 170 |
| Beam search | p. 170 |
| Observation reordering based pruning (ORBP) | p. 171 |
| Experiments | p. 171 |
| Experimental setup | p. 171 |
| Experimental results | p. 172 |
| Conclusions | p. 175 |
| Word Graph Based Feature Enhancement for Noisy Speech Recognition | p. 177 |
| Introduction | p. 177 |
| Word Graph Based Feature Enhancement | p. 178 |
| System Overview | p. 178 |
| Signal Processing Based Speech Enhancement | p. 179 |
| First Pass Decoding and Word Graph Construction | p. 179 |
| Model Based Clean Speech Synthesis | p. 180 |
| Wiener Filtering and Constrained Second Pass Decoding | p. 181 |
| Experiments | p. 182 |
| Experimental Setup | p. 182 |
| Signal Processing Based Speech Enhancement | p. 182 |
| GMM Based Feature Enhancement | p. 183 |
| Word Graph Based Feature Enhancement | p. 183 |
| Conclusions | p. 184 |
| References | p. 184 |
| Search and Mining | |
| Building Bridges for Web Query Classification | p. 189 |
| Introduction | p. 189 |
| Problem Definition | p. 191 |
| Query and Category Enrichment | p. 191 |
| Enrichment through Search Engines | p. 192 |
| Word Matching Between Categories | p. 192 |
| Classification Approaches | p. 192 |
| Classification by Exact Matching | p. 193 |
| Classification by SVM | p. 193 |
| Our New Method: Classifiers by Bridges | p. 194 |
| Experiments | p. 197 |
| Data Set and Evaluation Metrics | p. 197 |
| Results and Analysis | p. 198 |
| Related Work | p. 202 |
| Conclusion and Future Work | p. 202 |
| References | p. 203 |
| Spectral Domain-Transfer Learning | p. 205 |
| Introduction | p. 205 |
| Preliminaries on Spectral Methods | p. 207 |
| Cross-Domain Spectral Classification | p. 207 |
| Problem Definition | p. 207 |
| Objective Function | p. 207 |
| Incorporating Constraints | p. 209 |
| Optimization | p. 210 |
| Case Study | p. 212 |
| Experiments | p. 212 |
| Data Sets | p. 213 |
| Evaluation Metric | p. 215 |
| Comparison Methods | p. 215 |
| Implementation Details | p. 215 |
| Experimental Results | p. 216 |
| Related Work | p. 218 |
| Spectral Methods | p. 218 |
| Transfer Learning | p. 219 |
| Conclusion and Future Work | p. 220 |
| References | p. 220 |
| Table of Contents provided by Ingram. All Rights Reserved. |