Record Details
 
  « New Search    
   
 
Bibliographic Data
Control Number315408
Date and Time of Latest Transaction20160315064426.AM
General Information160315s |||||||||b ||00|||
Cataloging SourceSTII-DOST
Local Call Number(T) QA278.5 S67 1997
Main Entry - Personal NameSoriano, Maricor Narvaez
Title StatementImproved classification of noisy microscopic images by a neural network with principal component projections as inputs by Maricor Narvaez Soriano
Physical Description86 leaves figures, graphs, tables
Summary, Etc.The error propagation in a trained sigmoid-activated neural network with a principal components preprocessors (PCAN+NN is derived and is shown to be smaller than that of a trained neural network without a preprocessor (NN). Both networks are tested to identify graylevel 10x10 pixel images of methaphase spreads, unburst cells and empty slides corrupted with additive and multiplicative Gaussian noise, impulse noise, constant gain and additive and multiplicative Gaussian noise with gain. Results indicate that a PCAN+NN is indeed more robust than a NN in classifying noisy images. It is also shown that reconstruction of a noisy image using a small number of principal components improves image quality
Subject Added Entry - Topical TermPhysics
 Image processing -- Digital techniques
 Neural networks (Neurobiology)
 Nuclear reactors -- Noise
 Principal components analysis
LocationDOST STII (T) QA278.5 S67 1997 THESES T STI-13-2261 1 14-16902 Donation 2013-05-20
  
     
 
Physical Location
Department of Science and Technology
Science and Technology Information Institute(T) QA278.5 S67 1997
  
     
 
Digital Copy
Not Available
  
     
  |  Download MARC record Download MARC
 
         
         
Online Catalog
Basic Search
Advanced Search
Browse Subjects
Book Cart
 
         

Text Size:
S  -  M  -  L
Copyright © 2004-2024. Philippine eLib Project
Host: U.P. Diliman University Library