Arc-support Line Segments Revisited: An Efficient High-quality Ellipse Detection.

Area of Science:

• Computer Vision
• Image Processing
• Pattern Recognition

Background:

• Ellipse detection in real-world images remains a significant challenge despite numerous algorithms.
• Existing methods often struggle with accuracy and efficiency in complex visual data.

Purpose of the Study:

• To develop an industry-oriented ellipse detector that achieves high accuracy and efficiency.
• To address the limitations of current ellipse detection techniques in practical applications.

Main Methods:

• Extraction of arc-support line segments to simplify image curves while preserving essential properties.
• Iterative and robust linking of arc-support line segments to form groups belonging to common ellipses.
• Complementary fitting approaches (local saliency and global search) for rapid initial ellipse estimation.
• Hierarchical clustering in a 5D parameter space for formulating ellipse candidates.
• Stringent verification for refining salient ellipse candidates into final detections.

Main Results:

• The proposed method demonstrates high detection accuracy and efficiency.
• Achieved superior F-measure scores compared to state-of-the-art methods on three public datasets.
• The approach effectively handles complex curves and real-world image conditions.

Conclusions:

• The novel arc-support line segment-based ellipse detector offers a valuable solution for industry applications.
• The method provides a robust and efficient means for accurate ellipse detection.
• The open-source code facilitates further research and application development.