E.J. van Henten, G.E.H. Marx, J.W. Hofstee, J. Hemming, V. Sarlikioti
For a better understanding of growth and development of tomato plants in three dimensional space, tomato plants were monitored using a computer vision system. It is commonly known that leaves of tomato plants do not have a fixed position and orientation during the day; they move in response to changing environmental conditions such as the position of the sun. For better understanding, it was desired to quantify this motion. Using a stereovision concept, two cameras were mounted in an experimental greenhouse a short distance apart from each other to enable depth measurement. Markers were placed on strategic spots on the tomato plant branches and leaves in the field of view of both cameras. Images were taken every ten minutes during daytime on several consecutive days. In the greenhouse, a virtual 3D coordinate system was defined and camera and tomato plant position and orientation were defined in this coordinate system. Image processing techniques were used to trace the markers and the 3D position coordinate of each marker in each image was calculated to obtain the course of a marker during several days. Stems, branches, and leaf nerves were considered as kinematic mechanical, robot like, links and corresponding theory was used to model and calculate the motion of stems and leaves of a tomato plant. Analysis of the images showed both small (1-2 degrees) and large rotations (10 degrees or more) of the branches and the different leaves on a branch during the course of a day. Leaves on one side of a branch showed a parallel motion in the same direction; the leaves on the opposite side of the branch showed a mirrored motion. However, deviating patterns occurred too. The developed method proved to be able to precisely quantify the motion of stems, branches and leaves of tomato plants during several days.
van Henten, E.J., Marx, G.E.H., Hofstee, J.W., Hemming, J. and Sarlikioti, V. (2012). MEASURING LEAF MOTION OF TOMATO BY MACHINE VISION. Acta Hortic. 952, 915-922
DOI: 10.17660/ActaHortic.2012.952.116
stereo vision, image processing, transpiration, VPD, plant modelling

Acta Horticulturae