Fast and accurate circle detection using gradient-direction-based segmentation.

Area of Science:

• Computer Vision
• Image Processing
• Computational Geometry

Background:

• Accurate circle detection is crucial in various image analysis applications.
• Existing methods like Hough transform and gradient pair vectors have limitations in speed and reliability.

Purpose of the Study:

• To introduce the first fitting-based circle detection algorithm, named Fast and Accurate Circle (FACILE) detection algorithm.
• To evaluate FACILE's performance against other state-of-the-art circle detection methods.

Main Methods:

• Developed a novel algorithm based on gradient-direction-based edge clustering and direct least square fitting.
• Segmented edges, validated arcs, and merged them for accurate circle extraction.
• Implemented the algorithm in C++ for performance evaluation.

Main Results:

• FACILE demonstrated significantly superior processing speed and detection reliability compared to randomized Hough transform, standard Hough transform, and fast circle detection using gradient pair vectors on simulated data.
• On standard datasets, FACILE achieved a 35% higher recognition rate than robust and precise circular detection, a state-of-the-art arc detection method.
• FACILE also showed significant improvements in processing speed over existing methods.

Conclusions:

• FACILE is a highly effective and efficient algorithm for circle detection.
• The fitting-based approach offers substantial advantages in both accuracy and speed for image analysis tasks.