Premium
An Artificial Neural Network‐Based Method to Identify Five Classes of Almond According to Visual Features
Author(s) -
Teimouri Nima,
Omid Mahmoud,
Mollazade Kaveh,
Rajabipour Ali
Publication year - 2016
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.12255
Subject(s) - artificial intelligence , pattern recognition (psychology) , computer science , artificial neural network , support vector machine , feature vector , thresholding , principal component analysis , classifier (uml) , color space , segmentation , computer vision , image (mathematics)
The quality evaluation is one of the key factors that have a major impact on the final price of agricultural products. Nowadays, image processing‐based techniques are becoming as an acceptable and widespread in quality evaluation procedures. In this study, we develop a robust method based on image processing and computational intelligence for quality grading and classification of almonds. The images of five classes of almond including normal almond ( NA ), broken almond ( BA ), double almond ( DA ), wrinkled almond ( WA ) and shell of almond ( SA ) were acquired by a scanner. For segmentation of images, both H component in HSI color space and O tsu's thresholding method were applied. In the next step, the feature vector, which includes 8 shape features, 45 color features and 162 texture features, was composed. For choosing correlated and superior features among all the 215 extracted features, sensitivity analysis was applied. Principal component analysis method was also used to reduce the dimension of the feature vector. The classification of almonds into different classes was carried out by artificial neural networks ( ANNs ). Among different ANN structures, the 18‐7‐7‐5 topology was the most optimum classifier. The accuracy of ANN classifier for each class was 98.92% for NA , 99.46% for BA , 98.38% for DA , 98.92% for WA and 100% for SA . The technique can readily be extended for online sorting machines. Practical Applications One of the applications of this method is in the design and fabrication of real‐time grading and sorting machines. The biggest advantage of the presented algorithm is its high precision. The developed classifier is able to detect and eject defected almonds (broken, double, wrinkled and shell of almonds) out of a stream of almonds in the sorting process line. Therefore, if the processing time of the method is improved further, it can readily be used in an online sorting machine.