Visualizing plant responses: Novel insights possible through affordable imaging techniques in the greenhouse

Authors: Conley, Matthew; Hejl, Reagan; Serba, Desalegn; Williams, Clinton

Submitted to: Sensors
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/10/2024
Publication Date: 10/17/2024
Citation: Conley, M.M., Hejl, R.W., Serba, D.D., Williams, C.F. 2024. Visualizing plant responses: Novel insights possible through affordable imaging techniques in the greenhouse. Sensors. 24(20). Article 6676. https://doi.org/10.3390/s24206676.
DOI: https://doi.org/10.3390/s24206676

Interpretive Summary: Due to climate pressures and human resource demands, measurements of plant performance need to be affordable, accessible, and efficient. This could be possible utilizing modern camera technology which allows for changes in plant status over time. To assess this opportunity, turfgrass plants were photographed inside a custom box designed with uniform light and using an inexpensive camera. Images were then processed with common software for plant cover area, multiple color aspects, and sensitivity to image corrections. Findings were compared to spectral reflection data and other plant performance metrics including visual quality, biomass production, and water use. Promising results indicate that (color) RGB-based imagery with simple controls is sufficient to measure the effect of plant treatments, along with notable correlations with multiple metrics. This study demonstrates the potential of consumer-grade photography to capture plant phenotypic traits. However, this research underscores the need for further research to support image corrections standardization and better connect image data to biological processes.

Technical Abstract: Global climatic pressures and increased human demands create a modern necessity for efficient and affordable plant phenotyping unencumbered by arduous technical requirements. The analysis and archival of imagery become easier as modern camera technology and computers are leveraged. This facilitates the detection of vegetation status and changes over time. Using a custom lightbox, an inexpensive camera, and common software, turfgrass pots were photographed in a greenhouse environment over an 8-week experimental period. Subsequent imagery was analyzed for area cover, color metrics, and sensitivity to image corrections. Findings were compared to active spectral reflection data and previously reported measurements of visual quality, productivity, and water use. Results indicate that RGB-based (color) imagery with simple controls is sufficient to measure the effect of plant treatments. Notable correlations were observed for corrected imagery, including between percent yellow color area classification segment (%Y) with human visual quality ratings (VQ) (R = -0.89), dark green color index (DGCI) with clipping productivity in mg d-1 (mg) (R = 0.61), and an index combination term (COMB2) with water use in mm d-1 (mm) (R = -0.60). The calculation of green color cover area (%G) correlated with normalized difference vegetation index (NDVI) (R = 0.91) and its Red spectral component (R = -0.87). A CIELAB ratio (BA) correlated with normalized difference red-edge index (NDRE) (R = 0.90), and its Red-Edge (R = -0.74) spectral component, while a new calculation termed HSVi correlated strongest to the near-infrared (NIR) (R = 0.90) reflectance spectra. Additionally, COMB2 significantly differentiated between the treatment effects of date, mowing height, deficit irrigation, and their interactions (p < 0.001). Sensitivity and statistical analysis of typical image file formats and corrections that included JPEG (JPG), TIFF (TIF), geometric lens correction (LC), and color correction (CC) were conducted. Results underscore the need for further research to support standardization and better connect image data to biological processes. This study demonstrates the potential of consumer-grade photography to capture plant phenotypic traits.