Please use this identifier to cite or link to this item: http://repositorio.unicamp.br/jspui/handle/REPOSIP/107482
Type: Artigo de periódico
Title: Digit Recognition Applied To The Automation Of Evaluation Of Medical Equipment [reconhecimento De Dígitos Aplicado à Automação De Ensaios Em Equipamentos Medicos]
Author: da Cruz Alexandrino J.
Hermini A.H.
Lepikson H.A.
de Araujo Kalid R.
Leite H.J.D.
Abstract: This paper presents the development of an application for capturing an image of numeric displays, recognizing the digits contained therein and transferring them to an automation system that manages metrological performance tests on different types of medical equipment. Using a webcam and the Delphi® development tool, an algorithm was created based on segmentation of the digits and counting of the points contained in each segment. This counting results in a numerical set, a kind of signature of the digit. The software has the phases of training and recognition, as well as an evaluation mode. Three tests were applied to evaluate the performance of the application. At first, ten thousand different values were generated and recognized using three webcams of different manufacturers. In the second, using three different equipments, there was the recognition of 100 different readings and the hit rate verified by visual comparison. In the third, the application was set for recognizing the temperature indicated on the display of a neonatal incubator and controlled through the communication interface between applications, by a prototype version of a management system of medical equipment assays. Tests conducted in the first way resulted in a hit rate of 100% for two webcams and 99.96% for a third. In other tests the hit rate was 100%, demonstrating that the application is feasible and safe.
Editor: 
Rights: aberto
Identifier DOI: 10.4322/rbeb.2011.010
Address: http://www.scopus.com/inward/record.url?eid=2-s2.0-82555182730&partnerID=40&md5=c54c5632e005e0b0851438223c281e36
Date Issue: 2011
Appears in Collections:Unicamp - Artigos e Outros Documentos

Files in This Item:
File SizeFormat 
2-s2.0-82555182730.pdf960.38 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.