Abstract
There is an increasing interest in using 3D computer vision in precision agriculture. This calls for better quantitative evaluation and understanding of computer vision methods. This paper proposes a test framework using ray traced crop scenes that allows in-depth analysis of algorithm performance and finds the optimal hardware and light source setup before investing in expensive equipment and field experiments. It was expected to be a valuable tool to structure the otherwise incomprehensibly large information space and to see relationships between parameter configurations and crop features. Images of real plants with similar structural categories were annotated manually for comparison in order to validate the performance results on the synthesised images. The results showed substantial correlation between synthesized and real plants, but only when all error sources were accounted for in the simulation. However, there were exceptions where there were structural differences between the virtual plant and the real plant that were unaccounted for by its category. The test framework was evaluated to be a valuable tool to uncover information from complex data structures.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Christensen LK (2004) NPK deficiencies discrimination by use of spectral and spatial response. PhD Thesis, The Royal Veterinary and Agricultural University, Agro Technology Section, Department of Agricultural Sciences, Denmark
Fusiello A, Roberto V, Verri A (2000) Symmetric Stereo with Multiple Windowing. Intern J Pattern Recognit Artif Intel 14(8):1053–1066
Jimenez AR, Ceres R, Pons JL (2000) A survey of computer vision methods for locating fruit on trees. Trans ASAE 43(6):1911–1920
Kolmogorov V, Zabih R (2002) Multi-camera scene reconstruction via graph cuts. In: Heyden A, Sparr G, Nielsen M, Johansen P (eds) Proceedings of 7th European conference on computer vision, IEEE, 3, Springer Berlin/Heidelberg, pp. 82–96
Lee WS, Slaughter DC, Giles DK (1999) Robotic weed control system for tomatoes. Precision Agric 1(1):95–113
Lund I, Søgaard HT (2004) Robotic weeding–Plant recognition and micro spray on single weeds. 2004 In: Proceedings 5th international conference on microprocessor systems in agriculture. Jasinski B (ed) ISBN 83-915188-8-4, Plock, Poland, pp. 89–96
McBratney A, Whelan B, Ancev T, Bouma J (2005) Future directions in precision agriculture. Precision Agric 6(1):7–23
Mulligan J, Isler V, Daniilidis K (2002) Trinocular stereo: a new algorithm and its evaluation. Intern J Comput Vis 47(1–3):51–61
Mulligan J, Zabulis X, Kelshikar N, Daniilidis K, (2004) Stereo-based environment scanning for immersive telepresence. IEEE Trans Circuit Syst Video Technol 14(3):304–320
Nielsen M, Andersen HJ, Slaughter DC, Giles DK (2004) Detecting leaf features for automatic weed control using trinocular stereo vision. In: Mulla DJ (ed) Proceedings of the 7th international conference on precision agriculture, University of Minnesota, St. Paul, MN, USA, CD-ROM
Nielsen M, Andersen HJ, Slaughter DC, Granum E (2005a) Ground truth evaluation of 3d computer vision on non-rigid biological structures. In: Stafford J (ed) Proceedings of the 5th European conference on precision agriculture, Wageningen Academic Publishers, The Netherlands, pp. 549–556
Nielsen M, Andersen HJ, Granum E, (2005b) Comparative study of disparity estimations with multi-camera configurations in relation to descriptive parameters of complex biological objects. In: Hellwich O, Niini I, Ressl C, Rodehorst V, Scharstein D, Sturm P (eds) BenCOS–Towards benchmarking automated calibration, orientation and surface reconstruction from Images (ICCV 2005), ISPRS Working groups, Dresden, Germany, 63–68
Scharstein D, Szeliski R (2002) A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. Intern J Comput Vis 47(1–3):7–42
Scharstein D, Szeliski R (2003) High-accuracy stereo depth maps using structured light. In: Proceedings of 2003 IEEE computer society conference on Computer Vision and Pattern Recognition (CVPR), 1, Martin D (ed) IEEE, California, pp. 195–202
Snyder DL, Helstrom CW, Lanterman AD, Faisal M, White RL (1995) Compensation for readout noise in CCD images. J Opt Soc Am A Opt Image SciVis 12(2):272–283
Sun C (1999) Multi-resolution stereo matching using maximum-surface techniques. In: Proceedings of 5th digital image computing: techniques and applications, Perth, Australia, pp.195–200
Szeliski R, Zabih R (1999) An experimental comparison of stereo algorithms. In: Triggs B, Zisserman A, Szeliski R (eds) Proceedings of international workshop on vision algorithms, Springer, Berlin Heidelberg, pp. 1–19
Acknowledgements
Authors would like to thank the Danish Technical Research Council, the Danish Agricultural and Veterinary Research Council and the Danish Ministry of Food, Agriculture and Fisheries for funding through the National Research programme “Sustainable Technology in Agriculture”.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Nielsen, M., Andersen, H.J., Slaughter, D.C. et al. Ground truth evaluation of computer vision based 3D reconstruction of synthesized and real plant images. Precision Agric 8, 49–62 (2007). https://doi.org/10.1007/s11119-006-9028-3
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11119-006-9028-3