A Classifiable Method of Soft-Drink Crown-Cap Types Using Rotational Angles and Neural Network

Thumbnail Image
File(s)
Author(s)
Other Contributor(s)
University of the Thai Chamber of Commerce. School of Engineering
Publisher(s)
University of the Thai Chamber of Commerce
Date Issued
2007
ISSN
1906-1625
Resource Type
Text::Journal::Journal article
Language
English
Abstract
This paper presents a classifiable method ofSDCC types using rotational angles and backpropagationneural network (BPNN). In this method, a dimension imageof the SDCC types is suitably decreased by image resizing,and then they are converted to binary images (0, 1) forinputs of neural network (NN). Finally, the NN is classifiedinto the SDCC types or the others. In the experiment, 10crown caps (CCs) per a type of SDCC (4 types: Pepsi,Sprite, Greenspot and Coca-Cola) are trained by BPNNwhich an angle of each rotation for a CC is 5 degrees tomake a full 360 degree and therefore rotational images ofthe CC are totally 72 pictures, as well as 50 samples of eachtype are tested. The results reveal that an accuracy of theclassification of the four SDCC types is approximately97.5 % whilst the others are rejected.
Subject(s)
Journal
UTCC Engineering Research Papers
Access Rights
Open access
Rights
This work is protected by copyright. Reproduction or distribution of the work in any format is prohibited without written permission of the copyright owner.
Rights Holder
University of the Thai Chamber of Commerce
Bibliographic Citation
Adisorn Leelasantitham (2007) A Classifiable Method of Soft-Drink Crown-Cap Types Using Rotational Angles and Neural Network. UTCC Engineering Research Papers.
URI