Premium
Introducing new shape features for classification of cucumber fruit based on image processing technique and artificial neural networks
Author(s) -
Kheiralipour Kamran,
Pormah Abbas
Publication year - 2017
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.12558
Subject(s) - artificial neural network , artificial intelligence , preprocessor , computer science , centroid , pattern recognition (psychology) , image processing , matlab , classifier (uml) , feature extraction , sorting , homogeneity (statistics) , software , machine vision , computer vision , image (mathematics) , machine learning , algorithm , programming language , operating system
Shape‐based classification of fruits and vegetables is one of the most important applications of image processing and machine vision technology in post‐harvest processing of agricultural products. In this research, desirable (cylindrical), and undesirable (curved and conical) shapes of cucumber fruit were considered to be intelligently detected using image processing technique and artificial neural networks method. A new algorithm was programed for preprocessing and extraction of shape features from the images in MATLAB 2010a software. Beside common features, two new features including “centroid non homogeneity” and “width non homogeneity” were introduced and extracted. After feature selection, different neural network models were evaluated to classify the useful features. The best classifier model had accuracy of 97.1% with 4‐20‐2 structure. Practical applications The present research introduces new features to distinguish cucumber shape: desired (cylindrical) and undesired. The approach can be used to develop a sorting system based on machine vision to separate cucumber fruits to two classes in industry.