One Two
Sie sind hier: Startseite Research NoViSys - Ertragsschätzung im Weinbau

NoViSys

Novel viticulture systems for sustainable production and products

Traditional grapevines are highly susceptible to fungal infestation and require a frequent application of fungicides. Sustainable viticulture therefore needs fungus resistant cultivars with a reduced demand for fungicide application. New grapevine cultivars named PIWIs (pilzwiderstandsfähige Rebsorten) have already been created and new cultivation systems like the SMPH (semi-minimal pruned hedge) emerge. These new approaches need to be examined to gain an objective perspective of their potential by collecting phenotypic data. 

 PIWI

 

On the organ level, the most important phenotypic trait is the yield. It is defined as the crop weight per vine whereby the weight is dependent on the variation of the number of bunches per vine (60%), the number of berries per bunch (30%) and the berry size (10%). The true biological yield, described as weight per vine, is therefore statistically forecast through assessing the so called yield parameters: bunch weight, number of bunches, number of berries per bunch and berry weight.

SMPH_VSP

 

In viticulture this phenotypic data is traditionally collected directly in the field via visual and manual means by an experienced person. This approach is time consuming, subjective and prone to human errors. In the last years research therefore focuses strongly on developing automated and non-invasive sensor based methods to reduce labor costs, increase data acquisition speed and enhance measurement accuracy and objectivity.

Crucial steps on this way are the

a)      automation of data acquisition with one or several sensors,

b)      the automated interpretation of the data for future measurement tasks and

c)      the automated measurement of plant parameters important for phenotyping using the interpreted data.

 

 

A track driven vehicle called PHENObot consisting of a camera system, a real-time-kinematic GPS system for positioning as well as hardware for image storage and acquisition is used to visually capture an SMPH row fully with georeferenced RGB-images. An image was taken every 20cm along the path in three camera heights, thus acquiring multi-perspective images of the same object. 

Phenobot 

 

 

In a 1st post-processing step these images were used within a Multi-View-Stereo software to reconstruct a textured 3D point cloud of the whole trellis row. The effect of occlusion inherent in 2D phenotyping is reduced tremendously through the multi-perspectivity, thus delivering a more complete representation of the whole trellis row.

Row_detail 

row_zoom

 

 

A classification algorithm is then used in a 2nd step to automatically classify the raw point cloud data into the semantic plant components grape bunches and canopy using Matlab® and open source software. Automation further reduces the need for manual intervention by a human operator. 

classify

 

 

In a 3rd step the number of grape bunches, the number of berries and the berry size are determined using the classification results. In this way a large amount of objective and precise phenotypic data is generated.

grape_origberry_count

 

References: 

NoViSys Homepage 
Plant2030
Julius-Kühn Institut

Contacts:

Artikelaktionen