Simple Search:

Development Of An Optical Character Recognition Function System For Integrated Circuit Label Classification Using Neural Network


Citation

Mariappan, Vasan (2008) Development Of An Optical Character Recognition Function System For Integrated Circuit Label Classification Using Neural Network. Masters thesis, Universiti Putra Malaysia.

Abstract / Synopsis

Presently, many Integrated Circuit (IC) manufacturers are applying machine vision solution to ensure the legibility of characters printed on the top surface of IC Package. In template matching technique there are about 10% of ICs rejected due to very little defects in quality of marking even though the characters are correct. The objective of this project is to develop an IC inspection system that has optical character recognition function system by using neural network. Feed forward back propagation neural network method is used in this task. The system developed is able to read 36 characters ( A to Z and 0 to 9) printed on ICs. The recognition time in template matching is 650μs. In neural network technique, by feeding-in Raw Data, Feature, and Hybrid (combination of Raw Data and Feature), they clock 18.22μs, 15.64μs and 19.32μs respectively. The recognition accuracy is 96.26% for the former and 98.25%, 98.83% and 99.61% for the latter. This is a solution to minimise rejects of ICs in manufacturing process. The reduction of processing time in manufacturing process contributes to the increase of productivity. Moreover, application of this technique gives a solution to avoid mismatch of parts (ICs) in manufacturing lots.


Download File

[img]
Preview
PDF
FK_2008_16.pdf

Download (324kB)

Additional Metadata

Item Type: Thesis (Masters)
Subject: Neural computers
Subject: Optical pattern recognition
Call Number: FK 2008 16
Chairman Supervisor: Associate Professor Megat Mohamad Hamdan Megat Ahmad, PhD
Divisions: Faculty of Engineering
Depositing User: Nurul Hayatie Hashim
Date Deposited: 09 Apr 2010 10:13
Last Modified: 27 May 2013 15:22
URI: http://psasir.upm.edu.my/id/eprint/5396
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item