Potential for machine vision of grain crop features for nitrogen assessment

Potential for machine vision of grain crop features for nitrogen assessment

Existing approaches for determining nitrogen (N) requirements typically involve measuring biomass and sensing near-infrared-based crop reflectance indices. There is potential for automated assessments of tiller counts, plant size and colour using machine vision to help indicate plant N status. Existing demonstrations of machine vision systems are typically for a single field rather than multiple fields. A barley and wheat field study has been conducted to identify robustness of machine vision across multiple sites for assessing biomass, and plant N status and concentration. Three N trial sites were established in Western Australia and South Australia during the 2020 season with low and rich-N strips. Each strip and the paddock were sampled in five to seven locations for plant N uptake, plant N concentration, and plant response using crop dry biomass and machine vision cameras. Machine vision algorithms were implemented on oblique images to extract indicators of vigour (colour) and physical size (line length and density that represent tillers and branches). Linear regression analysis identified that a normalised green red difference index from the colour machine vision system was strongly correlated with biomass and could add value to biomass and plant N assessment. Further work is to incorporate machine vision parameters into a data-driven N decision making method.

Files associated with this item cannot be displayed due to copyright restrictions.