Automatic thermal image filtering for determining representative canopy temperatures in vineyards

J. Luus, D.N.J. Els, C. Poblete-Echeverría
Canopy temperature has been proven to be an accurate indicator of plant water stress in vineyards by using the crop water stress index (CWSI). The process of calculating representative canopy temperatures from thermal images is relevant as there are several other elements in the images such as soil, sky, and other non-leaf materials that need to be excluded. In this study, five automatic methods of determining a representative canopy temperature were evaluated. The methods include: i) simple average, ii) temperature histogram analysis, iii) interval averaging, iv) colour thresholding with the RGB images, and v) reference temperature thresholding with Twet and Tdry measurements. Thermal images taken in a commercial vineyard on six days, with varying climatic conditions, were used to evaluate the methods. The automatic methods were evaluated against accurate manually analysed images in which only leaf material was carefully selected. The results show that the simple average method does not provide accurate temperatures. Colour thresholding does not perform as well as expected. The results indicate that the reference temperature thresholding technique, histogram analysis and interval analysis can calculate canopy temperatures within 1°C of the manually analysed images. The reference temperature thresholding technique does require accurate measurements of Twet and Tdry, therefore, histogram method (peak detection) and interval averaging are preferred for its simplicity and accuracy.
Luus, J., Els, D.N.J. and Poblete-Echeverría, C. (2024). Automatic thermal image filtering for determining representative canopy temperatures in vineyards. Acta Hortic. 1390, 285-292
DOI: 10.17660/ActaHortic.2024.1390.35
thermography, crop water stress index, temperature thresholding

Acta Horticulturae