Enhanced Vineyard Descriptors combining UAV 2D and 3D Crop Models

Precision viticulture has been assumed as an essential approach to optimise crop-managing practices and to improve the quality of food products. To deploy proper site-specific management, addressing the intrinsic variability within a vineyard or a parcel, reliable methods for features extraction and mapping of crops must be developed. The introduction in agriculture of Unmanned Aerial Vehicle (UAV), equipped with sensors able to acquire fields planar images at several wavelengths, makes available huge amount of data with high-resolution, in terms of both spatial and temporal dimension. Recently, in addition to well-known 2D mosaicked images, innovative features leaded by modern photogrammetry allowed accurate three-dimensional models of crops (ex. 3D point cloud datasets) to be generated. The approach presented in this work is aimed at defining enhanced crop descriptors by exploiting information provided by both 2D and 3D crop models. Crucial phases of the procedure are the proper management of data provided by several sources, in order to achieve high consistency of the obtained huge dataset. In addition, the detection of vine rows, discriminating them from all the other elements of rural areas, plays a crucial role. The proposed methodology does not require the straightness of vine rows and it can be profitably applied to models of vineyards with curvilinear rows, also on steep terrains. Specific computing optimisation have been defined in order to reduce big data complexity. A set of 24 portion of vine rows, each made by 4 plants, has been used to validate the effectiveness of the evaluated crop canopy descriptors. The 2D maps and 3D point-cloud models have been generated by using aerial images acquired during UAV flights at 35 meters high, in a study vineyard located in Serralunga d’Alba (Piedmont, Northwest of Italy). The integration of 2D-3D information allowed to obtain good performance also in the presence of dense inter-row grassing which, usually, slightly differs from vine canopies in terms of reflectance.

[1]  Lorenzo Comba,et al.  Vineyard detection from unmanned aerial systems images , 2015, Comput. Electron. Agric..

[2]  Alessandro Matese,et al.  NDVI-based vigour maps production using automatic detection of vine rows in ultra-high resolution aerial images , 2015 .

[3]  Won Suk Lee,et al.  Sensing systems for precision agriculture in Florida , 2015, Comput. Electron. Agric..

[4]  John P. Fulton,et al.  An overview of current and potential applications of thermal remote sensing in precision agriculture , 2017, Comput. Electron. Agric..

[5]  Carlos Lopes,et al.  Thermal data to monitor crop-water status in irrigated Mediterranean viticulture , 2016 .

[6]  Marco Bietresato,et al.  Evaluation of a LiDAR-based 3D-stereoscopic vision system for crop-monitoring applications , 2016, Comput. Electron. Agric..

[7]  Fabio Terribile,et al.  A geospatial decision support system for supporting quality viticulture at the landscape scale , 2017, Comput. Electron. Agric..

[8]  N. Zhang,et al.  Precision agriculture—a worldwide overview , 2002 .

[9]  Mark Whitty,et al.  Non-Productive Vine Canopy Estimation through Proximal and Remote Sensing , 2016 .

[10]  David Hernández-López,et al.  Vineyard yield estimation by automatic 3D bunch modelling in field conditions , 2015, Comput. Electron. Agric..

[11]  Lorenzo Comba,et al.  Individual plant definition and missing plant characterization in vineyards from high-resolution UAV imagery , 2017 .

[12]  Bernhard Höfle,et al.  Comparative classification analysis of post-harvest growth detection from terrestrial LiDAR point clouds in precision agriculture , 2015 .

[13]  Santiago Tosetti,et al.  Optimization methodology to fruit grove mapping in precision agriculture , 2015, Comput. Electron. Agric..

[14]  M. Srbinovska,et al.  Environmental parameters monitoring in precision agriculture using wireless sensor networks , 2015 .

[15]  Volker Steinhage,et al.  High-precision 3D detection and reconstruction of grapes from laser range data for efficient phenotyping based on supervised learning , 2017, Comput. Electron. Agric..

[16]  Frédéric Baret,et al.  Using 3D Point Clouds Derived from UAV RGB Imagery to Describe Vineyard 3D Macro-Structure , 2017, Remote. Sens..

[17]  Jinha Jung,et al.  Crop height monitoring with digital imagery from Unmanned Aerial System (UAS) , 2017, Comput. Electron. Agric..

[18]  Giulio Reina,et al.  Terrain assessment for precision agriculture using vehicle dynamic modelling , 2017, ArXiv.

[19]  Pablo J. Zarco-Tejada,et al.  Tree height quantification using very high resolution imagery acquired from an unmanned aerial vehicle (UAV) and automatic 3D photo-reconstruction methods , 2014 .

[20]  Anders Krogh Mortensen,et al.  Weight prediction of broiler chickens using 3D computer vision , 2016, Comput. Electron. Agric..

[21]  Elling Ruud Øye,et al.  A 3D machine vision system for quality grading of Atlantic salmon , 2016, Comput. Electron. Agric..

[22]  Ke Wang,et al.  LSSA_CAU: An interactive 3d point clouds analysis software for body measurement of livestock with similar forms of cows or pigs , 2017, Comput. Electron. Agric..

[23]  Hong Sun,et al.  Potato feature prediction based on machine vision and 3D model rebuilding , 2017, Comput. Electron. Agric..

[24]  Pablo J. Zarco-Tejada,et al.  Grape quality assessment in vineyards affected by iron deficiency chlorosis using narrow-band physiological remote sensing indices , 2010 .

[25]  Simon Bennertz,et al.  Combining UAV-based plant height from crop surface models, visible, and near infrared vegetation indices for biomass monitoring in barley , 2015, Int. J. Appl. Earth Obs. Geoinformation.

[26]  Lorenzo Comba,et al.  Unsupervised detection of vineyards by 3D point-cloud UAV photogrammetry for precision agriculture , 2018, Comput. Electron. Agric..