[en] This paper presents a 3D stereo-vision system aiming at the measurement of plant
characteristics. The method is intended to be used in an Ecotron where the availability of
plant material is limited and where crop should be characterised non destructively. The plants
were not considered as individuals, it was the crop as a whole that was characterised. The
leaf area index (LAI) was measured. It is an important property of vegetation, since it
determines the photosynthetic primary production, the plant evaporation and characterises the
plant growth. The average leaf angle (ALA) was also measured. For the computation, leaf
pixels were differentiated from soil pixels by using linear discriminant analysis. The stereo
vision system computed the distance to the camera of each pixel in the image in the region
where the pixels are present in both images. The observed area was computed on the basis of
the average distance of the leaf pixels in the region. The leaf area was evaluated for each
triplet of adjacent pixels by computing the cross product of the vectors defined by those three
points. The sum gave the leaf area for the same region. The area of these triangles was
summed for all the pixels in the region and the ratio to the observed area gave the LAI. The
ALA was the mean orientation of the pixel triplets. After calibration, the method was found
to present a coefficient of correlation of 0.93 with destructive reference measurements and a
precision of 0.12 for the LAI. It was possible to measure the LAI and the ALA from the
germination up to the ripening stage with a minimum of work load.