ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

UAV-Based Multispectral Imaging for Discriminating Cover Crops and Weeds at the Species Level

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2026 ASABE Annual International Meeting  2600379.(doi:10.13031/aim.202600379)
Authors:   Shekhar Suman Borah, Prabha Sundaravadivel, Vijay Varanasi, Krishna Reddy, Reginald Fletcher, Prasanna Gowda
Keywords:   Cover Crop Classification, Curly Dock, Italian Ryegrass, Multispectral Imaging, Normalized Difference Vegetation Index (NDVI), Precision Agriculture, Unmanned Aerial Vehicle (UAV).

Abstract. Accurate species-level discrimination in mixed vegetation fields is essential for site-specific weed management and cover crop stand assessment. Crimson clover (Trifolium incarnatum L.) is widely used as a winter cover crop but often coexists with aggressive weeds such as Italian ryegrass (Lolium perenne ssp. multiflorum) and curly dock (Rumex crispus L.), making manual scouting labor-intensive and error prone. UAV-based RGB and multispectral imaging provides a scalable, non-destructive alternative for field-level monitoring. This study evaluates the ability of low-altitude UAV multispectral imagery to discriminate among crimson clover, Italian ryegrass, and curly dock under field conditions. High-resolution RGB and multispectral data were collected at 3.96 m altitude over an USDA experimental field in Stoneville, Mississippi. Vegetation indices including Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), and Normalized Difference Red Edge Index (NDRE) were computed and integrated with GLCM texture features, HSV-based color segmentation, and morphological shape metrics to extract species-specific spectral and structural signatures. In addition, close-range RGB imagery was analyzed using RGB and HSV color spaces, visible-band vegetation indices (ExG, NGRDI, GLI), texture features, and morphological descriptors to provide ground-level validation. Distinct spectral and morphological patterns were observed across species. Crimson clover exhibited the highest mean NDVI (0.72 ± 0.14), strong near-infrared reflectance, and characteristic red flower clusters detectable through HSV segmentation. Italian ryegrass was effectively identified using GNDVI (0.61 ± 0.12) combined with elongated shape features, while curly dock showed the lowest NDRE values (0.42 ± 0.09), reflecting altered red-edge behavior. Texture analysis indicated that green-band contrast and dissimilarity were particularly effective for delineating fine leaf structures. This approach provides a foundation for precision agriculture tools that support targeted weed management, reduce herbicide inputs, and improve cover crop stand evaluation.

(Download PDF)    (Export to EndNotes)