Sun Jun
School of Electrical and Information Engineering of Jiangsu University, 212013, Zhenjiang, People Republic of China
Jiang Shuying
School of Electrical and Information Engineering of Jiangsu University, 212013, Zhenjiang, People Republic of China
Mao Hanping
Laboratory Venlo of Modern Agricultural Equipment, Jiangsu University, 212013, Zhenjiang, People Republic of China
Zhang Xiaodong
Laboratory Venlo of Modern Agricultural Equipment, Jiangsu University, 212013, Zhenjiang, People Republic of China
Zhu Wenjing
Laboratory Venlo of Modern Agricultural Equipment, Jiangsu University, 212013, Zhenjiang, People Republic of China
Wang Yan
School of Electrical and Information Engineering of Jiangsu University, 212013, Zhenjiang, People Republic of China
ABSTRACT
The feature extraction and optimization of lettuce leaf image are the important premise of classification recognition of lettuce nitrogen levels. The lettuce samples of different nitrogen levels were cultivated in soilless cultivation using nitrogen nutrition of different concentrations. When the lettuce leaf images were collected, image features have been extracted, including texture features, shape features and color features. Because of the redundancy of characteristic values, there were influences in the accuracy and efficiency of image recognition. Genetic algorithm was used to optimize 11 eigenvalues and the Principal Component Analysis (PCA) dimension reduction method was used to choose 12 principal component feature values whose cumulative contribution rate reached 98.24%. Later, the Support Vector Machine (SVM) was used as classifier. The 90 samples were chosen as training samples and the remaining 30 samples were chosen as the test samples. The result shows that, the prediction accuracy of SVM classifier based on genetic algorithm feature optimization reaches 93.33% and that based on PCA features optimization reaches 76.67%. So the genetic algorithm feature optimization is more suitable for lettuce leaf image feature optimization.
PDF References Citation
How to cite this article
Sun Jun, Jiang Shuying, Mao Hanping, Zhang Xiaodong, Zhu Wenjing and Wang Yan, 2013. Classification of Lettuce Nitrogen Levels Based on Image Feature Extraction and Optimization. Information Technology Journal, 12: 7574-7579.
DOI: 10.3923/itj.2013.7574.7579
URL: https://scialert.net/abstract/?doi=itj.2013.7574.7579
DOI: 10.3923/itj.2013.7574.7579
URL: https://scialert.net/abstract/?doi=itj.2013.7574.7579
REFERENCES
- Laddi, A., S. Sharma, A. Kumar and P. Kapur, 2013. Classification of tea grains based upon image texture feature analysis under different illumination conditions. J. Food Eng., 115: 226-231.
CrossRefDirect Link - Bazi, Y. and F. Melgani, 2007. Semisupervised PSO-SVM regression for biophysical parameter estimation. IEEE Trans. Geosci. Remote Sensing, 45: 1887-1895.
CrossRefDirect Link - Chen, B.J., H.Z. Shu and H. Zhang, G. Chen, C. Toumoulin, J.L. Dillenseger and L.M. Luo, 2012. Quaternion zernike moments and their invariants for color image analysis and object recognition. Signal Proces., 92: 308-318.
CrossRef - Guan, H.O., S.H. Xu and F. Tan, 2010. Image segmentation model of plant lesion based on genetic algorithm and fuzzy neural network. Trans. Chinese Soc. Agric. Machinery, 41: 163-167.
Direct Link - Luo, G.H., W.M. Huang and L. Song, 2010. 2-D maximum entropy spermatozoa image segmentation based on canny operator. Proceedings of the International Conference on Intelligent Computing and Integrated Systems, October 22-24, 2010, Guilin, pp: 243-246.
CrossRef - Ning, Y.Y., W.J. Li and X.N. Wang, 2010. Method for vehicle-logo recognition based on principal components analysis and BP neural network. J. Liaoning Normal Univ., 2: 179-184.
Direct Link - Sharma, A. and A. Kuh, 2008. Class document frequency as a learned feature for text categorization. Proceedings of the International Joint Conference on Neural Networks, June 1-8, 2008, Hong-Kong, China, pp: 2988-2993.
CrossRef - Sun, J., Y. Wang, X. Wu, X. Zhang and H. Gao, 2012. A new image segmentation algorithm and its application in lettuce object segmentation. Telkomnika, 10: 557-563.
Direct Link