Kubacki, Jens: Learning and detecting objects in combined range and color images based on local feature frames. 2012
Inhalt
- Acronyms
- Notation
- Introduction
- Related Work
- Classical Vision and Recognition Paradigms
- Newer Approaches to Robot Vision
- Algorithms for Object Recognition
- Pure Learning Approaches
- Modern Invariant Feature Point Detectors
- Feature Points and Classifiers
- Geometric Models and Matching
- Combined 3D and 2D-based Approaches
- Integration 3D Imaging Sensors
- Summary
- Problem and Approach
- General Motivation
- Classes of Object Recognition
- Observations and Inspirations
- Rationale on Algorithm Level
- System Overview
- Summary
- Sensor Fusion
- Sample Data Acquisition
- Motivation and Approach
- Range Segmentation
- Possible Acquisition Scenarios
- Range Segmentation Examples
- Summary
- Feature Frame Cloud Extraction and Matching
- Motivation and Approach
- Feature Frame Cloud Computation
- Discrete Descriptor Key Computation
- Clustering Feature Point Descriptors
- Local Frame Construction
- Example
- Model Construction and Detection
- Summary
- Evaluation
- Combined Sensor
- FFC Computation and Matching
- Descriptor Tests
- System Tests
- Learning and Detection with Known Learning Frames
- Learning and Detection with Unknown Learning Frames
- Live Tests
- Summary
- Summary and Conclusions
- Appendix
- Glossary