This study investigates robust keypoint detection and geometric image matching in high-resolution UAV imagery of agricultural fields – an essential component for precision farming, crop monitoring, yield prediction, and automated field boundary mapping. While unmanned aerial vehicle (UAV) systems provide high spatial resolution and flexibility, aligning multiple images into coherent mosaics remains a technical challenge, particularly in agricultural settings where repetitive structures, low texture, and illumination variations are prevalent. Feature-based approaches like ORB and SIFT have been widely adopted in remote sensing and photogrammetry, yet their effectiveness in such field-specific conditions is still insufficiently characterized. This paper aims to fill that gap by evaluating both methods under controlled scenarios using UAV images captured at two altitudes, employing Lowe’s ratio test and RANSAC-based homography estimation for validation.
ORB, a lightweight algorithm based on FAST keypoints and BRIEF descriptors, was tested under three configurations varying the number of features, pyramid levels, and scale factors. The results reveal that ORB struggles to extract reliable features in low-contrast or repetitive farmland scenes, often yielding insufficient inliers despite parameter optimization. SIFT, on the other hand, utilizes multi-octave scale-space analysis and gradient-based descriptors to detect stable, rotation- and scale-invariant keypoints. A comprehensive grid search was conducted to fine-tune SIFT’s n_features, ratio_threshold, and ransac_threshold, resulting in a configuration that achieved 100% inlier ratio and reduced false matches significantly.
The findings highlight SIFT’s superior robustness and reliability in complex agricultural image alignment tasks. Despite its higher computational cost, its descriptive power ensures accurate registration, especially in structurally repetitive or low-texture environments. This study contributes practical insights into algorithmic trade-offs between efficiency and accuracy, and offers a validated SIFT+RANSAC pipeline with tuning guidelines for UAV-based agricultural mosaicking. These results may support future hybrid solutions that integrate classical and deep learning-based feature detectors for scalable, field-ready applications.
1. Yu, Z., Zhou, H., & Li, C. (2017). Fast non-rigid image feature matching for agricultural UAV via probabilistic inference with regularization techniques. [Conference paper].
2. Rublee, E., Rabaud, V., Konolige, K., & Bradski, G. (2011). ORB: An efficient alternative to SIFT or SURF. In Proceedings of the IEEE International Conference on Computer Vision (ICCV).
3. Dibs, H., Idrees, M., Saeidi, V., & Mansor, S. (2016). Automatic keypoints extraction from UAV image with refine and improved scale invariant features transform (RI-SIFT). [Conference paper].
4. Fesiuk, A., & Furgala, Y. (2023). Keypoints on the images: Comparison of detection by different methods. Electronics and Information Technologies.
5. Lowe, D. G. (2004). Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision.
6. Lingua, A., Marenchino, D., & Nex, F. (2009). Performance analysis of the SIFT operator for automatic feature extraction and matching in photogrammetric applications [Journal article].
7. Kang, P., & Ma, H. (2011). An automatic airborne image mosaicing method based on the SIFT feature matching [Conference paper].
8. Rublee, E., Rabaud, V., Konolige, K., & Bradski, G. (2011). ORB: An efficient alternative to SIFT or SURF. In Proceedings of the IEEE International Conference on Computer Vision (ICCV).
9. Baid, U. (2015). Image registration and homography estimation [Technical report].
10. Pankaj, D. S., & Nidamanuri, R. R. (2016). A robust estimation technique for 3D point cloud registration [Journal article].
11. Zhang, X., Tian, Y., Zhu, Y., et al. (2019). Rapid mosaicking of UAV images for crop growth monitoring using the SIFT algorithm [Journal article].
12. EOS Crop Monitoring (n. d.). Main map: Fields. Retrieved from https://crop-monitoring.eos.com/main-map/fields/all
13. Mur-Artal, R., Montiel, J. M. M., & Tardós, J. D. (2015). ORB-SLAM: A versatile and accurate monocular SLAM system [Journal article].
14. Dibs, H., Idrees, M., Saeidi, V., & Mansor, S. (2016). Automatic keypoints extraction from UAV image with refine and improved scale invariant features transform (RI-SIFT) [Conference paper].
15. Alcantarilla, P. F., Nuevo, J., & Bartoli, A. (2013). Fast explicit diffusion for accelerated features in nonlinear scale spaces. In Proceedings of the British Machine Vision Conference (BMVC).
16. Leutenegger, S., Chli, M., & Siegwart, R. Y. (2011). BRISK: Binary robust invariant scalable keypoints. In Proceedings of the IEEE International Conference on Computer Vision (ICCV).
17. DeTone, D., Malisiewicz, T., & Rabinovich, A. (2011). SuperPoint: Self-supervised interest point detection and description [Conference paper].
18. Bergstra, J., & Bengio, Y. (2012). Random search for hyper-parameter optimization. Journal of Machine Learning Research.