Estimation of the height and angles of orientation of the upper leaves in the maize canopy using Stereovision

Abstract

3D models of vegetation have been constructed to allow various simulations, such as the interception of incoming radiation, radiative transfer, reflectance simulation in remote sensing, study of plant function, and interspecies competition. The construction of such 3-D models is based upon simple geometric assumptions and several production rules (eg, growing, branching). It poses 2 problems: the acquisition of a large amount of data on leaf heights and orientation angles in the field; and the validation of the model and assessment of its suitability on the scale of the field. This paper presents a method of 3D acquisition and reconstruction based upon Stereovision, ie upon the acquisition of data using 2 synchronized CCD or photographic cameras. We first show that Toscani's method of camera calibration can be used under real conditions in the field. We then apply the method to a tall maize canopy (2.50 m high) because this canopy allows easy control measurements to be made in the field. Results on the estimation of position and orientation of the leaves are presented. This method allows the estimation of the X, Y and Z coordinates of characteristic points of laminae and stems. The ability of the method to provide inclination angles of the leaves was evaluated to be rather good. Only the leaves belonging to the 4 upper layers were reconstructed because of the difficulties due to hidden areas. This problem is discussed and we show how auxilliary information on crop structure may be complementary to data obtained from Stereovision. © 1994.