Real-time non-invasive leaf area measurement method using depth images

Otoya P.E.L.; Gardini S.R.P. · 2020 IEEE ANDESCON, ANDESCON 2020 (2020) · DOI: 10.1109/andescon50619.2020.9271993

Resumen

In this work, we present a new method for estimating the surface area of wavy edged single leaves, through obtaining triangular meshes from depth images. First, a set of Pelargonium Hortorum leaves are imaged (one by one) using a stereo camera for obtaining the depth information between a leaf and the camera, which is expressed in depth image format. Second, depth images are binarized in order to obtain the connected component that contains only the pixels that define the leaf surface image. Third, a triangular mesh is built as a representation of the leaf surface, integrating the information of the connected component, the depth image and the intrinsic parameters of the camera. Finally, the estimated leaf surface area is calculated as the sum of the areas of all the obtained triangles. The algorithm developed here, allowed the triangulation of a set of 40 000 points in less than a second.

Instituciones: Universidad Privada Antenor Orrego