Related Experiment Videos
Convergence condition and efficient implementation of the fuzzy curve-tracing (FCT) algorithm
1Department of Computer Engineering and Information Technology, City University of Hong Kong, Kowloon, Hong Kong. h.yan@cityu.edu.hk
Summary
The fuzzy curve-tracing (FCT) algorithm reliably extracts smooth curves from noisy data. This study proves FCT convergence and introduces methods for faster, more stable curve tracing, outperforming snake models.
Area of Science:
- Computer Vision
- Image Processing
- Computational Geometry
Background:
- Extracting smooth curves from noisy, unordered data is challenging.
- Existing methods like snake models have limitations in accuracy and speed.
Purpose of the Study:
- To analyze the convergence properties of the fuzzy curve-tracing (FCT) algorithm.
- To develop enhanced methods for fast, reliable, and stable FCT implementation.
- To demonstrate FCT's superiority over conventional boundary detection techniques.
Main Methods:
- Analysis of FCT convergence based on matrix diagonal dominance.
- Development of initialization, multiresolution clustering, and image down-sampling techniques.
- Implementation of methods to avoid common curve-tracing errors.
Main Results:
- Convergence of FCT is proven under specific weighting coefficient conditions.
- Improved algorithm stability and speed achieved through proposed methods.
- Demonstrated advantages of FCT over snake models for curve extraction and boundary detection.
Conclusions:
- The fuzzy curve-tracing algorithm offers a robust and efficient solution for extracting smooth curves from noisy data.
- The developed implementation methods significantly enhance FCT's practical applicability.
- FCT presents a valuable alternative to existing methods in image analysis and data processing.