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, Click here for options.

Development of a Robust Weed Species Mapping System using Hyperspectral Imaging for Precision Weed Control in Processing Tomato

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

Citation:  2010 Pittsburgh, Pennsylvania, June 20 - June 23, 2010  1009313.(doi:10.13031/2013.29850)
Authors:   Y Zhang, D C Slaughter
Keywords:   Hyperspectral imaging, weed mapping, machine learning, plant identification, diurnal temperature.

Californian processing tomatoes are planted over a 4-month period, during which there is a substantial change in the growing environment. Most of the prior research using visible and near infrared spectroscopy for plant identification have been limited in scope to plants grown under a single environmental condition. This work studied the impacts of variations in diurnal growing temperature range on hyperspectral features in the visible and near infrared region with a specific focus on species identification and high-resolution mapping. Six major Californian processing tomato cultivars, black nightshade (Solanum nigrum L.) and red-root pigweed (Amaranthus retroflexus L.) were grown under a variety of diurnal temperature ranges simulating the range common in the Californian springtime planting period and one treatment simulating greenhouse growing conditions. The principal change in canopy reflectance with changing temperature occurred in the 480 670 nm and 695 795 nm regions. A species-temperature interaction effect between tomato and black nightshade was observed in the performance of a species classifier based upon the full 400795 nm range. A bilaterally symmetric adverse effect on the classification rates of these two species occurred when the classification models were applied to plants grown under a diurnal temperature range different from the conditions under which the training plants were grown. The overall species classification rate was higher for models (91%) trained under mild conditions compared to models (83%) trained under higher diurnal temperature range conditions and were nearly as high as the classification rate (92%) achieved by models calibrated at global temperature conditions.

(Download PDF)    (Export to EndNotes)