ENSEEIHTENSEEIHT

VINNEO is a project which brings together companies and labaratories for wine production quality improvement in the south-west of France. Our contribution is to compute some useful caracteristics of the grapeyard, e.g. the discontinuity in the foliage, the leaves density or the average size of a vine on a parcel. Our goal is thus to infere these data from a visual analysis. In order to do that, we have acquired videos passing through the rows of the vineyard. A first step has been to segment the foliage using color and depth informations. Using these data, and assumption about plant models and vineyards, we apply a new skeletonisation step to extract the branching structure. The last step is the plant modeling using L-systems.

This work has lead to several publications :

ISVC 2013: Reconstructing Plants in 3D from a Single Image using Analysis-by-Synthesis (Jérôme Guénard, Géraldine Morin, Frédéric Boudon and Vincent Charvillat) accepted for oral presentation, pdf , presentation

abstract: Mature computer vision techniques allow the reconstruction of challenging 3D objects from images. However, due to high complexity of plant topology, dedicated methods for generating 3D plant models must be devised. We propose to generate a 3D model of a plant, using an analysis-by-synthesis method mixing information from a single image and a priori knowledge of the plant species. First, our dedicated skeletonisation algorithm generates a possible branching structure from the foliage segmentation. Then, a 3D generative model, based on a parametric model of branching systems that takes into account botanical knowledge is built. The resulting skeleton follows the hierarchical organisation of natural branching structures. An instance of a 3D model can be generated. Moreover, varying parameter values of the generative model (main branching structure of the plant and foliage), we produce a series of candidate models. The reconstruction is improved by selecting the model among these proposals based on a matching criterion with the image. Realistic results obtained on different species of plants illustrate the performance of the proposed method.

Curves and Surfaces 2012: Realistic Plant Modeling from Images based on Analysis-by-Synthesis (Jérôme Guénard, Géraldine Morin, Frédéric Boudon and Vincent Charvillat) pdf , presentation

abstract: Plants are essential elements of virtual worlds to get pleasant and realistic 3D environments. Even if mature computer vision techniques allow the reconstruction of challenging 3D objects from images, due to high complexity of plant topology, dedicated methods for generating 3D plant models must be devised. We propose an analysis-by-synthesis method which generates 3D models of a plant from both images and a priori knowledge of the plant species. Our method is based on a skeletonisation algorithm which allows to generate a possible skeleton from a foliage segmentation. Then, a 3D generative model, based on a parametric model of branching systems that takes into account botanical knowledge is built. This method extends previous works by constraining the resulting skeleton to follow hierarchical organisation of natural branching structure. A first instance of a 3D model is generated. A reprojection of this model is compared with the original image. Then, we show that selecting the model from multiple proposals for the main branching structure of the plant and for the foliage improves the quality of the generated 3D model. Varying parameter values of the generative model, we produce a series of candidate models. A criterion based on comparing 3D virtual plant reprojection with original image selects the best model. Realistic results obtained on different species of plants illustrate the performance of the proposed method.

ORASIS 2011: Reconstruction de modèles virtuels de vignes à partir d'images (Jérôme Guénard, Géraldine Morin, Frédéric Boudon, Pierre Gurdjos and Vincent Charvillat) pdf , presentation

abstract: We propose a method for reconstructing virtual model of vines from images. For this, an analysis by synthesis method is used and consist in characterizing an image using a number of a priori knowledge about the 3D scene. Initially, we get an approximation of the plant modeled in 3D. Then, comparing its reprojection with the original image, we refine this model though an iterative optimisation process. To be efficient, our method do not optimize positionning of individual leaves but rather a realistic foliage consistent with the images.

AFIG 2010: Modélisation de vignes à partir d'une séquence d'images (Jérôme Guénard, Charlotte Giron, Géraldine Morin, Frédéric Boudon, Pierre Gurdjos and Vincent Charvillat) pdf , presentation

abstract: Cet article présente des travaux sur la modélisation de plantes à géométries fortement contraintes à partir d'images. A partir de séquences d'images acquises dans un vignoble, nous instancions un modèle paramétré des parcelles, des rangs, et des pieds de vignes. Le modèle est déduit des connaissances a priori ; à partir des images, des paramètres sont extraits. Ces paramètres sont ensuite fournis au modèle qui génère une représentation de la plante, du rang ou de la parcelle filmée.