Dr. Julio César Pastrana Pérez














Fachgebiet Biosystem- und Gartenbautechnik
Leibniz Universität Hannover
Herrenhäuser Straße 2
D-30419
E-Mail: pastrana (at) bgt.uni-hannover.de
Telefon: +49 (0) 511 - 762 - 3621
Fax: +49 (0) 511 - 762 2649



 

Research
Robotic vision is very important when developing robots that will autonomously perform farming or horticultural tasks. The capacity of a robotic system to cope with the production demands very much depends on how accurately and efficiently it recognizes the objects while performing its task.

 

Occluded or partially occluded objects are commonly found when trying to identify different plants or crops. Therefore, it is very important to develop vision systems that are not susceptible to occlusions and overlapping.

 

Currently, a method to detect small plants under overlapping situations is being developed. This technique is based on detecting single leaves for further grouping into individual plants. The leaf detection is carried out by detecting ellipses that best superimpose the shape of a plant, considering that each leaf has to a certain degree an elliptical form. After detecting all possible leaves (ellipses) the next step is to group them into single plants by using an Active Shape Model deformable template.

 

Active shape modeling is also being used to identify the best shape model for soil pores. The pore 3D structures were generated by analyzing tomographic images that were produced with a synchrotron radiation source. It is worth noting that the voxel length in the x, y and z direction is about 700 nm.

 

Publications

 

 
Doctoral thesis: Active shape models with focus on overlapping problems applied to plantdetection and soil pore analysis. Leibniz Universittät Hannover. 2012.
J. Pastrana

 

 
Role of Fungal Mycelium in the Formation of Carbonate Concretions in Growing Media - An Investigation by SEM and Synchrotron-Based X-Ray Tomographic Microscopy.
S. Masaphy, L. Zabari, J. Pastrana, S. Dultz. Geomicrobiology Journal, vol. 29, no. 7, pp. 442-450. 2009.

 

 
3-dimensional vision analysis of highly resolved tomography images of fine pore systems of soil and growing media samples.
J. Pastrana, T. Rath, S. Dultz. International conference on agricultural engineering, AgEng, Hersonissos, Crete, Greece, 23-25 June 2008.

 

 
Vision system identification of overlapping plants.
J. Pastrana, T. Rath. Greensys, International Symposium for High Technology for Greenhouse System, Quebec, Canda. 14.-19.6.2009.

 

 
3-D-Bildverarbeitung von röntgenmikrotomografischen Bodenanalysen im Mikrometerbereich.
J. Pastrana, T. Rath, S. Dultz. Bornimer Agrartechnische Berichte, Heft 62, 24-30. Leibniz-Institut für Agrartechnik Potsdam-Bornim, 2008. ISSN 0947-7314.

 

 
Vision based plant recognition under overlapping situations.
J. Pastrana, T. Rath. Bornimer Agrartechnische Berichte, Heft 62, 16-23. Leibniz-Institut für Agrartechnik Potsdam-Bornim, 2008. ISSN 0947-7314.

 

 
Vision based plant recognition under overlapping situations in horticulture.
J. Pastrana, T. Rath. First Symposium on Horticulture in Europe. University of Vienna, 17-20.02. 2008, SHE, S. 164, ISSN 1996-9449, ISHS. (Poster)

 

M. Bennewitz, J. Pastrana, and W. Burgard. In: Proceedings of the Third Workshop on Selforganization of Adaptive Behavior (SOAVE), 2004

 

Julio Pastrana, University of Freiburg, 2005.

 

 
 
 
Impressum                   (c) BGT, 2012