Crop Detection and Positioning in the Field Using Discriminant Analysis and Neural Networks Based on Shape Features

Authors
S. Kiani
A. Jafari
Department of Agricultural Machinery, College of Agriculture, Shiraz University, Shiraz, Islamic Republic of Iran.
Abstract
Development of an autonomous weeding machine requires a vision system capable of detecting and locating the position of the crop. It is important for the vision system to be able to recognize the accurate position of the crop stem to be protected during weeding. Several shape features of corn plants and common weed species in the location were extracted by means of morphological operations. Effective features in the classification of corn and weeds were analyzed using stepwise discriminant analysis. Among the seven features used in the analysis, four were sufficient to classify the two target groups of weeds and corn. These shape features were fed to artificial neural networks to discriminate between the weeds and the main crop. 180 images consisting of corn plants and four species of common weeds were collected from normal conditions of the field. Results showed that this technique was able to distinguish corn plants with an accuracy of 100% while at most 4% of the weeds were incorrectly classified as corn. In the final stage, the position of the main crop was also approximated and its accuracy was measured with respect to the real position of the crop. The position of the crop is necessary for the weeding machine to root up all of the plants except the main crop. It was concluded that the high accuracy of this method is due to the significant difference between corn and weeds in the critical period of weeding in the region.

Keywords

1. Ahmad U. and Kondo, N. 1997. Weed Detection in Lawn Field. http://mama.agr.okayama-u.ac.jp/lase/weed.htm .
2. Astrand, B. and Baerveldt, A. J. 2003. Mobile Robot for Mechanical Weed Control. Int. Sugar J., 105(1250): 89-95.
3. Cho, S. I., Lee, D. S. and Jeong, J. Y. 2002. Weed-plant Discrimination by Machine Vision and Artificial Neural Network. Biosystems Eng., 83 (3): 275–280.
4. Jafari, A., Mohtasebi, S. S., Eghbaliand, H. and Omid, M. 2006. Weed Detection in Sugar Beet Fields Using Machine Vision. Int. J. Agric. Biol. 8(5): 602-605.
5. Meyer, G. E. Mehta, T., Kocher, M. F., Mortensen, D. A., Samal, A. 1998. Textural Imaging and Discrimanant Analysis for Distinguishing Weeds for Spot Spraying. Transactions of the ASAE, 41(4), 1189–1197.
6. Pan, J., Min, H. and Yong, H. 2007. Crop and Weed Image Recognition by Morphological Operations and ANN Model. Instrumentation and Measurement Technology Conference – IMTC. Warsaw, Poland, May 1-3, 2007.
7. Pe´rez, A. J., Lopez, F., Benlloch, J. V. and Christensen, S. 2000. Colour and Shape Analysis Techniques for Weed Detection in Cereal Felds. Comput. Electron. Agric. 25: 197–212.
8. Polder, G., Van Evert, F. K., Lamber, A. 2007. Weed Detection Using Textural Image Analysis. EFITA. http://www.efita.net/apps/accesbase/dbsommaire.asp?d=6257.
9. Woebbeck, D. M., Meyer, G. E., Von Bargen, K. and Mortensen, D. A. 1995. Color Indices for Weed Identification under Various Soil, Residue and Lighting Conditions. Trans. ASAE, 38(1): 259-269.
10. Zhu, B., Jiang, L., Luo, Y. and Tao, Y. 2007. Gabor Feature-based Apple Quality Inspection Using Kernel Principal Component Analysis. J. Food Eng. 8(4): 741-749.
Journal of Agricultural Science and Technology
Volume 14, Issue 4
July and August 2012
Pages 755-765