Abstract
We propose a new algorithm for real-time detection and tracking of elliptic patterns suitable for real-world robotics applications. The method fits ellipses to each contour in the image frame and rejects ellipses that do not yield a good fit. The resulting detection and tracking method is lightweight enough to be used on robots' resource-limited onboard computers, can deal with lighting variations and detect the pattern even when the view is partial. The method is tested on an example application of an autonomous UAV landing on a fast-moving vehicle to show its performance indoors, outdoors, and in simulation on a real-world robotics task. The comparison with other well-known ellipse detection methods shows that our proposed algorithm outperforms other methods with the F1 score of 0.981 on a dataset with over 1500 frames. The videos of experiments, the source codes, and the collected dataset are provided with the letter.
Introduction
Detecting circular or elliptical shapes is a fundamental task in robotics, essential for identifying landing pads for UAVs or road signs for autonomous vehicles. However, real-world environments present significant challenges, including motion blur, shadows, reflections, and partial occlusions. Standard techniques like the Hough Transform are often too computationally expensive for resource-constrained onboard computers, while optimization-based methods can produce excessive false positives.
This research introduces a lightweight, high-speed algorithm designed to bridge the gap between simple geometric methods and complex, resource-heavy approaches. By combining a least-squares fitting technique with rigorous geometric validation and dynamic image scaling, the proposed method achieves high accuracy while maintaining the real-time speeds necessary for autonomous operations.
Methodology
The proposed algorithm follows a "fit-then-validate" pipeline designed for efficiency:
- Edge and Contour Extraction: The system utilizes the Canny edge detector to group pixels into contours.
- Least-Squares Fitting: For every contour, an ellipse is fitted using the Fitzgibbon and Fisher method. This provides a fast initial estimate.
- Strict Validation: To filter out false detections, the algorithm applies several geometric tests, including dimension checks (rejecting unrealistic axis ratios) and contour/ellipse overlap checks. These checks ensure that the fitted ellipse is well-supported by actual image edges rather than random noise.
- Dynamic Tracking and Scaling: Once a target is identified, the search is narrowed to a Region of Interest (ROI). The system employs dynamic scaling: downsampling the image when the target is close to save CPU cycles, and scaling up to search the full frame if the target is lost.
This approach prioritizes fast validation over complex optimization, making it highly suitable for real-time robotics applications where the target may only be partially visible or obscured.
Experimental Results
The algorithm was evaluated using the AirLab Elliptic Target Detection Dataset, comprising 1,511 frames with various lighting conditions and pattern thicknesses. The performance was compared against established methods, including the Xie & Ji Hough-based approach and the arc-grouping method by Fornaciari et al.
The proposed method achieved an F1 score of 0.981, with a precision of 99.77% and a recall of 96.44%, significantly outperforming competitors. In terms of speed, the algorithm averaged 32.2 ms per frame on an Intel Core i5 mobile CPU. Notably, when the tracking ROI and dynamic scaling were active, the processing time dropped to just a few milliseconds, demonstrating its efficiency for onboard deployment.
Conclusion
The presented real-time ellipse detection algorithm offers a robust solution for autonomous robotics tasks, such as landing UAVs on moving platforms. Its ability to handle partial views and maintain high speeds on limited hardware makes it a practical choice for field deployment. Future work will focus on improving performance under extreme perspective skew and integrating more sophisticated tracking filters to enhance stability during aggressive maneuvers.
Summary updated