International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 1248
ISSN 2229-5518
India is the second largest producer of wheat in the world after China. Determining the quality of wheat is critical. Specifying the quality of wheat manually requires an expert judgement and is time consuming. Sometimes the variety of wheat looks so similar that differentiating them becomes a very tedious task when carried out manually. To overcome this problem, machine algorithms can be used to classify wheat according to its quality.
Machine algorithms are incorporated by using machine vision. Machine vision is widely used in the field of agriculture for identifying the varieties of various food crops and for identifying their quality as well. A machine vision system (MVS) provides an alternative to the manual inspection of biological products. Machine vision system incorporates the use of digital images. These images are obtained with the help of digital camera and are then stored in the computer for future work. In a Machine Vision System, the camera acts as an eye and the computer acts as the brain.
Meesha Punn is currently pursuing masters degree program in Computer Science in Punjab Technical University, Punjab India. E-mail:meesha990@yahoo.in
Nidhi Bhalla is currently working as Assistant
Professor in Swami Vivekanand Institute of
Engineering and Technology, Punjab, India.
Digital images stored in the computer are processed by Image processing algorithms extract the features from the digital images and use them to generate pattern. These patterns are input to the machine algorithms based on which the objects are classified into their respective classes. Such machine algorithms used for classifying objects are referred to as pattern classifiers.
Image processing is being increasingly applied in the very sensitive area seed analysis. It is also an excellent instrument for making tests and doing monitoring and classification in a number of other industrial production areas. Image processing applied to wheat seeds’ quality classification contributes in achieving fast and accurate operation.
The only drawback of machine vision system [10] is that its results are influenced by the quality of the image captured by the camera and accuracy of machine algorithm(s).
The methodology for classification of wheat grains incorporates the following steps:
IJSER © 2013 http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 1249
ISSN 2229-5518
Capture Wheat grain image
Image enhancement
Image segmentation
Feature extraction
Machine algorithm
Performance evaluation
noise and blurring from the image. Image enhancement is done to get better understanding of the image.
Image segmentation refers to the process of dividing a digital image into various segments. The purpose of segmentation is to present the image in more simplified presentation so that it is easy to analyze the image and get the segments of interest. Image segmentation gives a meaning to the image.
Selecting features deals with choosing relevant subset of features form large set of features to carry the task of classification. While classification of wheat, identification, extraction and selection of appropriate features is of great importance. This is so because selecting the wrong features can deviate the classification process from its correct path. The features can be used in
Fig 1. Procedure for wheat grain identification
The first step for the classification of wheat using machine algorithms is to acquire the images of wheat. The images are acquired using a digital camera and are stored in the computer in the form of digital images.
Image scaling is the process of resizing the images. Increasing the size of the image makes the image soft. Reducing the size of the image enhances the smoothness and sharpness of the image.
Image enhancement deals with modifying the image so as to improve the image. It improves the quality of image by removing
combination with other features in order to obtain the better accuracy of the machine learning algorithms[1]. Employment of correct number of features offers better results. Employing more features can deteriorate the performance of the machine algorithm and increase computational cost[7]. Correct number and selection of features leads to computational accuracy of machine algorithm.
IJSER © 2013 http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 1250
ISSN 2229-5518
Rectangularity | |
Roundness | |
Compactness | |
Length | |
Width | |
Major Axis Length | |
Minor Axis Length | |
Textural Features | Mean |
Textural Features | Standard Deviation |
Color Features | Hue |
Color Features | Saturation |
Color Features | intensity |
Table 1. Features extracted for Machine
Algorithms
Feature selection using SFS algorithm Sequential Forward Selection algorithm incorporates a criterion function based on scatter matrix[13]. It chooses feature that has a larger value in the criterion function and adds that function to vector function[1]. Feature selection using UTA algorithm UTA algorithm calculates the average of feature in all instances. This selected feature is replaced by calculated mean value in all input vectors. Then the trained network is tested with the new features and total feature’s error is calculated [3].
Different machine algorithms have been
implemented till date using different features of grains. An algorithm was developed taking into account morphological features to classify Canada Western Red Spring(CWRS) wheat, Canada Western Amber durum(CWAD) wheat, barley, oats and rye [4]. Another classification model was developed by combining two or three feature sets(morphological, color, textural) for classification of Canada Western Red Spring(CWRS) wheat, Canada Western Amber durum(CWAD) wheat, barley, oats and rye [5]. Different values of accuracy
was achieved using different pair of feature sets. Later, a Digital Image Analysis (DIA) algorithm was developed to classify Canada Western Red Spring(CWRS) wheat, Canada Western Amber durum(CWAD) wheat, barley, oats and rye [6] using textural features of individual wheat grains. Textural features were obtained using different color band combinations to identify the color band combination that classifies the wheat grains with maximum accuracy [6].
A comparison of three machine learning
classifiers [8], namely Artificial Neural Network (ANN), Support Vector Machine (SVM)[12] and Random Forest (RF) was done [9]. The comparison showed that RF took longer time than SVM but has an advantage that it is easy to use. This is so because it requires only one variable to be set by the user. ANN produced results between RF and SVM. It required the highest calculation times among the three classifiers. ANN was considered as the least favorable classifier. Conversely, SVM emerged as the best classifier. It showed good performance and robustness. The calculations were done faster and took less calculation time than ANN. Burges et al.,
1998 stated that Support Vector Machine (SVM) is a set of linear classifier that provides higher values of classification accuracy as compared to other classifiers used such as multilayer perceptron neural networks [14].
Another study employed ANN classifier for classification of wheat using two features, morphology and color [2]. It was concluded that morphological features were better than color features in classifying wheat. But, only morphological features were not sufficient in wheat classification so a combination of both was used.
After the implementation of machine
algorithms, the performance of classifiers is evaluated and the results are obtained.
IJSER © 2013 http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 1251
ISSN 2229-5518
Agricultural Produce (Grading & Marking) Act, 1937 is also referred as AGMARK Standards [17]. It provides standards of various agricultural commodities. Grading provides description of the quality of the wheat seeds. The grading standard deals with the following:
[15] wheat and (2) vulgare or common wheat [16].
Grade Designation | Foreign matter (% by wt.) | Other food grains (% by wt.) | Other wheats (% by wt.) | Damaged grains (% by wt.) | Slightly damaged grains (% by wt.) | Immature shrivelled and Broken Grains (% by wt.) | Weevilled Grains (% by wt.) |
I | 1.0 | 1.6 | 5.0 | 1.0 | 2.0 | 2.0 | 1.0 |
II | 1.0 | 3.0 | 15.0 | 2.0 | 4.0 | 4.0 | 3.0 |
III | 1.0 | 6.0 | 20.0 | 4.0 | 6.0 | 10.0 | 6.0 |
IV | 1.0 | 8.0 | 20.0 | 5.0 | 10.0 | 10.0 | 10.0 |
Table 2 List of Quality Parameters for Wheat Grain Classification
The classification accuracy varies differently for different classifiers. Also, the accuracy
varies differently for same classifier employing different set of feature values. A
single feature alone may not be enough for proper classification of wheat so a
combination of two or more features can be used to obtained better classification of
wheat. Employing large number of feature values does not improve the performance of machine algorithm. It increases the computational cost and may also decrease the accuracy of machine classifier. Choosing the correct features results in better accuracy of machine classifier. After comparison of various machine learning algorithms it was concluded that SVM emerged as the best
IJSER © 2013 http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 1252
ISSN 2229-5518
classifier. It showed good performance and robustness and took less computational time as compared to other machine learning algorithms.
[1] F.Guevara-Hernandez and J. Gomez-Gil
2011 “A machine vision system for
classification of wheat and barley grain kernels”.
[2] A.Arefi, A.Modarres Motlagh and R. Farrokhi Teimourlou 2011 “Wheat class
identification using computer vision system and artificial neural networks”.
[3] Alireza Pazoki and Zohreh Pazoki, 2011 “Classification system for rain fed wheat
cultivars using artificial neural network”.
[4] Majumdar, S. and D.S. Jayas. 2000a.
“Classification of cereal grains using machine vision. I. Morphology models.”
[5] Majumdar, S. and D.S. Jayas. 2000b. “Classification of cereal grains using
machine vision. II. Color models”.
[6] Majumdar, S. and D.S. Jayas. 2000c.
“Classification of cereal grains using machine vision. III. Texture models”.
[7] Majumdar, S. and D.S. Jayas. 2000d. “Classification of cereal grains using
machine vision. IV. Combined morphology, color, and texture models”.
[8] Francisco Pereira, Tom Mitchell, Matthew Botvinivick, 2008 “Machine
learning classifiers and fMRI: A tutorial overview”.
[9] I. Nitze, U. Schulthess, H. Asche, 2012 “Comparison of machine learning
algorithms Random Forest, Artificial Neural
Network and Support Vector Machine to
maximum likelihood for supervised crop type classification”.
[10] Narendra V G, Hareesh K S, 2010 “Quality inspection and grading of
agricultural and food products by computer vision- A review”.
[11] Sanjivani Shantaiya, Mrs.Uzma Ansari,
2010 “Identification Of Food Grains And Its
Quality Using Pattern Classification”.
[12] William S Noble, 2006 “What is a
support vector machine, Nature
Biotechnology”.
[13] Theodoridis S, Koutroumbas K, 20006 “Pattern recognition”.
[14] Burges, C. J. C. “A tutorial on support vector machines for pattern recognition”.
[15] B S Tyagi, Jag Shoran, Gyanendra
Singh, R K Gupta and B Mishra, 2008.
“Breeding durum wheat genotypes for better grain yield, pasta quality and resistance.”
[16] Dharam Pal, Sanjay Kumar, 2009. “
Genetic analysis of forage yield and other
traits in barley (Hordeum vulgare L.)” [17] http://agmarknet.nic.in/Wheat_manual.htm
IJSER © 2013 http://www.ijser.org