You need JavaScript to view this

Image processing applied to automatic detection of defects during ultrasonic examination; Imagerie numerique ultrasonore pour la detection automatique de defauts en controle non destructif

Abstract

This work is a study about image processing applied to ultrasonic BSCAN images which are obtained in the field of non destructive testing of weld. The goal is to define what image processing techniques can bring to ameliorate the exploitation of the data collected and, more precisely, what image processing can do to extract the meaningful echoes which enable to characterize and to size the defects. The report presents non destructive testing by ultrasounds in the nuclear field and it indicates specificities of the propagation of ultrasonic waves in austenitic weld. It gives a state of the art of the data processing applied to ultrasonic images in nondestructive evaluation. A new image analysis is then developed. It is based on a powerful tool, the co-occurrence matrix. This matrix enables to represent, in a whole representation, relations between amplitudes of couples of pixels. From the matrix analysis, a new complete and automatic method has been set down in order to define a threshold which separates echoes from noise. An automatic interpretation of the ultrasonic echoes is then possible. Complete validation has been done with standard pieces.
Authors:
Publication Date:
Oct 01, 1992
Product Type:
Thesis/Dissertation
Report Number:
CEA-R-5623
Reference Number:
SCA: 420500; PA: AIX-24:030510; SN: 93000956011
Resource Relation:
Other Information: TH: These (D. es Sc.).; PBD: Oct 1992
Subject:
42 ENGINEERING; AUSTENITIC STEELS; ULTRASONIC TESTING; IMAGE PROCESSING; MATRICES; ALGORITHMS; OPTIMIZATION; S CODES; ULTRASONIC WAVES; WAVE PROPAGATION; 420500; MATERIALS TESTING
OSTI ID:
10132923
Research Organizations:
Institut National des Sciences Appliquees (INSA), 69 - Villeurbanne (France); CEA Centre d`Etudes Nucleaires de Saclay, 91 - Gif-sur-Yvette (France). Direction des Technologies Avancees
Country of Origin:
France
Language:
French
Other Identifying Numbers:
Other: ON: DE93619562; TRN: FR9301841030510
Availability:
OSTI; NTIS (US Sales Only); INIS
Submitting Site:
FRN
Size:
[251] p.
Announcement Date:
Jul 04, 2005

Citation Formats

Moysan, J. Image processing applied to automatic detection of defects during ultrasonic examination; Imagerie numerique ultrasonore pour la detection automatique de defauts en controle non destructif. France: N. p., 1992. Web.
Moysan, J. Image processing applied to automatic detection of defects during ultrasonic examination; Imagerie numerique ultrasonore pour la detection automatique de defauts en controle non destructif. France.
Moysan, J. 1992. "Image processing applied to automatic detection of defects during ultrasonic examination; Imagerie numerique ultrasonore pour la detection automatique de defauts en controle non destructif." France.
@misc{etde_10132923,
title = {Image processing applied to automatic detection of defects during ultrasonic examination; Imagerie numerique ultrasonore pour la detection automatique de defauts en controle non destructif}
author = {Moysan, J}
abstractNote = {This work is a study about image processing applied to ultrasonic BSCAN images which are obtained in the field of non destructive testing of weld. The goal is to define what image processing techniques can bring to ameliorate the exploitation of the data collected and, more precisely, what image processing can do to extract the meaningful echoes which enable to characterize and to size the defects. The report presents non destructive testing by ultrasounds in the nuclear field and it indicates specificities of the propagation of ultrasonic waves in austenitic weld. It gives a state of the art of the data processing applied to ultrasonic images in nondestructive evaluation. A new image analysis is then developed. It is based on a powerful tool, the co-occurrence matrix. This matrix enables to represent, in a whole representation, relations between amplitudes of couples of pixels. From the matrix analysis, a new complete and automatic method has been set down in order to define a threshold which separates echoes from noise. An automatic interpretation of the ultrasonic echoes is then possible. Complete validation has been done with standard pieces.}
place = {France}
year = {1992}
month = {Oct}
}