Comparison of different classification algorithms for certain weed seeds species and wheat grains identification based on morphological parameters
The manual seeds identification takes a long time for practical applications. Therefore, automatic reliable plant seeds identification is of great technical and economic importance in agricultural industry. In this study, a system for automatic seed identification was developed based on optimal set of morphological features. The system uses digital images, which are processed to derive morphological features of the ten weed species seeds and wheat grains. The optimal morphological features are used for the discrimination of weed species seeds and wheat grains by using a classification step. Since the classification is a pivotal step for the seeds identification, four different classification algorithms namely K-Nearest Neighbours, Nave Bayes, Quadratic Discriminant Analysis and Feed-Forward Backpropagation Neural Network classifiers are evaluated to select the effective one. Among the four tested classifiers, Quadratic Discriminant Analysis classifier reported the highest identification accuracy of 97.1% and the minimum running time of 0.49 second.
Keywords: weed seeds; wheat grains; morphological features; identification; classification.
Download Full-Text
ABOUT THE AUTHORS
Enas Mohamed Kamel Mohamed
Enas Mohamed Kamel Mohamed received her B.Sc. degree in pure mathematics and computer sciences from Ain Shams University in 1997 and received her M.Sc. degree in Computer Sciences from Helwan University in 2009. Enas is working Assistant researcher at Weed Research Central Laboratory, Agriculture Research Center, Giza, Egypt.
Maged Hussein Wafy
Maged Hussein Wafy holds M.Sc. from Helwan University, and a Ph.D. degree in Uncertain Reasoning from Manchester University. He is Assistant Professor of the Department of Information Technology, Faculty of Computer and Information Systems, Helwan University. Currently, he works on the development of detection and identification of local features in images, visual cryptography, identification gait, machine learning, and Bayesian statistics and their applications in computer vision.
Hashem Mohamed Mohamed Ibrahim
Hashem Mohamed Mohamed Ibrahim received his B.Sc. Agriculture 1969, Vegetable Department, University of Alexandria, M.Sc. 1975, Horticulture Dept., Ain Shams University, PhD. 1981, Horticulture Dept. Ain Shams University. He is a professor of Weed Research Central Laboratory, Agriculture Research Center, Giza, Egypt.
Iman Asaad Badr Othman
Iman Badr received her PhD in Computer Engineering from Stuttgart University, Germany, and her M.Sc. and B.Sc. in Computer Science from the American University in Cairo, Egypt 1996 and from the Faculty of Computers and Information, Helwan University, Egypt, 2000, respectively. The topic of her PhD is "Agent-based dynamic scheduling for Flexible Manufacturing Systems". The topic of her M.Sc. thesis is agent-based information filtering for internet- based applications. Since 2011- present, lecturer at Helwan University, Egypt. 2006-2010: researcher and scientific employee at Stuttgart University, Germany. She is a member of the IEEE Industrial Electronics Society and the VDI/ VDE German Technical Committee on Multi-Agent Systems “FA 5.15”. Her research interests include agent-based computing, factory automation, operation research and bio-inspired optimization.
Enas Mohamed Kamel Mohamed
Enas Mohamed Kamel Mohamed received her B.Sc. degree in pure mathematics and computer sciences from Ain Shams University in 1997 and received her M.Sc. degree in Computer Sciences from Helwan University in 2009. Enas is working Assistant researcher at Weed Research Central Laboratory, Agriculture Research Center, Giza, Egypt.
Maged Hussein Wafy
Maged Hussein Wafy holds M.Sc. from Helwan University, and a Ph.D. degree in Uncertain Reasoning from Manchester University. He is Assistant Professor of the Department of Information Technology, Faculty of Computer and Information Systems, Helwan University. Currently, he works on the development of detection and identification of local features in images, visual cryptography, identification gait, machine learning, and Bayesian statistics and their applications in computer vision.
Hashem Mohamed Mohamed Ibrahim
Hashem Mohamed Mohamed Ibrahim received his B.Sc. Agriculture 1969, Vegetable Department, University of Alexandria, M.Sc. 1975, Horticulture Dept., Ain Shams University, PhD. 1981, Horticulture Dept. Ain Shams University. He is a professor of Weed Research Central Laboratory, Agriculture Research Center, Giza, Egypt.
Iman Asaad Badr Othman
Iman Badr received her PhD in Computer Engineering from Stuttgart University, Germany, and her M.Sc. and B.Sc. in Computer Science from the American University in Cairo, Egypt 1996 and from the Faculty of Computers and Information, Helwan University, Egypt, 2000, respectively. The topic of her PhD is "Agent-based dynamic scheduling for Flexible Manufacturing Systems". The topic of her M.Sc. thesis is agent-based information filtering for internet- based applications. Since 2011- present, lecturer at Helwan University, Egypt. 2006-2010: researcher and scientific employee at Stuttgart University, Germany. She is a member of the IEEE Industrial Electronics Society and the VDI/ VDE German Technical Committee on Multi-Agent Systems “FA 5.15”. Her research interests include agent-based computing, factory automation, operation research and bio-inspired optimization.