Local Scale-Invariant Contour Features for Object Recognition / Markus Hennig ; Erster Gutachter: Prof. Dr. Erdal Kayacan, Zweiter Gutachter: Prof. Dr. Reinhold Häb-Umbach. Paderborn, 2025
Inhalt
- Abstract
- Zusammenfassung
- Contents
- 1 Introduction
- 2 Background and Motivation
- 3 General Ambiguity Model for Binary Edge Images
- 3.1 Introduction
- 3.2 Related Work
- 3.3 Ambiguity Model and Tracing
- 3.3.1 Modeling Principles
- 3.3.2 Procedure and Pseudocode Implementation
- 3.3.3 Proof of Correctness
- 3.3.4 Simplified Implementation
- 3.3.5 Fundamental Test Cases
- 3.3.6 Application Examples
- 3.4 Evaluation
- 3.4.1 Dataset Construction
- 3.4.2 Method Characteristics Overview
- 3.4.3 Method Analysis
- 3.4.4 Runtime Analysis
- 3.5 Summary and Final Remarks
- 4 Local Scale-Invariant Contour Features
- 4.1 Scale-Space Theory
- 4.1.1 Problem Characterization
- 4.1.2 Scale-Space Representations
- 4.1.3 Scale-Space Axioms
- 4.1.4 Computational Implementation
- 4.2 Related Work
- 4.3 Keypoint Detection and Scale Assignment
- 4.3.1 Why Simple Methods are Insufficient
- 4.3.2 Curvature Scale-Space Construction
- 4.3.3 Detection of Curvature Extrema
- 4.3.4 Tracing of Curvature Extrema
- 4.3.5 Assigning Characteristic Scales
- 4.3.6 Assigning Characteristic Orientations
- 4.3.7 Extension to Open Contours
- 4.3.8 Box Filter Approximation
- 4.4 Evaluation
- 5 Conclusion and Future Work
- Bibliography
- List of Acronyms
- List of Figures
- List of Tables
- List of Symbols