EUVIMEDEUVIMED entdecken
Crossref

AgRowStitch: A Leaf-Scale Image Stitching Pipeline for Ground-Based Agricultural Images

Isaac K. Uyehara, Heesup Yun, Kyle T. Rizzo, Earl Ranario, Mason Earles

Applied Engineering in Agriculture · 2026 · 42 (3) · 307-317

DOI: 10.13031/aea.16650 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Applied Engineering in Agriculture, 2026. Verfasst von Isaac K. Uyehara und 4 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Highlights Current software struggles to make high-resolution image mosaics from agricultural images. We created a user-friendly and open-source pipeline for stitching agricultural images taken along a row of crops. AgRowStitch produced high-resolution mosaics across multiple sensors when other pipelines failed. The mosaic quality allows for crop monitoring and phenotyping of leaf-scale features. Abstract. Agricultural imaging often requires individual images to be stitched together into a final mosaic for analysis. However, agricultural images can be particularly challenging to stitch because the images contain similar objects, plants are neither rigid nor planar, and mosaics built from many images can accumulate errors that cause drift. Although these issues can be minimized by using images that cover a large area, this precludes the analysis of leaf-scale features. AgRowStitch was created as a user-friendly and open-source pipeline for stitching high-resolution images derived from proximal sensing. The AgRowStitch pipeline operates on single crop rows, so the geometry of the final mosaic can be used to constrain the stitching process. These geometric constraints are used to restrict the image-matching process, filter key point matches generated by SuperPoint and LightGlue, and straighten the mosaics after compositing. AgRowStitch was tested on images taken from three different ground-based imaging platforms. The mosaics produced by AgRowStitch were nearly seamless and had minimal artifacts, while the OpenCV and OpenDroneMap pipelines failed to produce any mosaics of the full region of interest, and Metashape only produced a mosaic of the full imaging area for one dataset. AgRowStitch was robust to small changes in camera orientation and operated without needing additional geospatial information. This makes AgRowStitch a general solution for creating row mosaics suitable for crop monitoring and phenotyping at scales ranging from individual leaves to flowers and fruits. Keywords: Metashape, Mosaic, OpenCV, OpenDroneMap, Panorama, Proximal sensing, Smartphone.

Zitieren

APA

Uyehara, I. K., Yun, H., Rizzo, K. T., Ranario, E., & Earles, M. (2026). AgRowStitch: A Leaf-Scale Image Stitching Pipeline for Ground-Based Agricultural Images. Applied Engineering in Agriculture, 42(3), 307-317. https://doi.org/10.13031/aea.16650

Vancouver

Uyehara IK, Yun H, Rizzo KT, Ranario E, Earles M. AgRowStitch: A Leaf-Scale Image Stitching Pipeline for Ground-Based Agricultural Images. Applied Engineering in Agriculture. 2026;42(3):307-317. doi:10.13031/aea.16650

BibTeX
@article{Uyehara2026,
  author = {Uyehara, Isaac K. and Yun, Heesup and Rizzo, Kyle T. and Ranario, Earl and Earles, Mason},
  title = {AgRowStitch: A Leaf-Scale Image Stitching Pipeline for Ground-Based Agricultural Images},
  journal = {Applied Engineering in Agriculture},
  year = {2026},
  volume = {42},
  number = {3},
  pages = {307--317},
  doi = {10.13031/aea.16650}
}
Kostenlos starten →