Location

Seattle, WA

Start Date

1-1-1996 12:00 AM

Description

NDT technologies by ultrasounds are in constant progress. The advance is considerable for sensors and acquisition systems. The means for data processing follow this evolution. In order to meet the demand for processing very high volumes of data, we propose in this paper a method of volume automatic thresholding. The application concerns the testing of austenitic steel or else, for the nuclear industry. The thresholding techniques from the image histogram are inadequate as the ultrasonic image histogram is unimodal. The study of the image with the cooccurrence matrix which is a two dimensionnal histogram allows to clearly show the noise-defect transition. Several authors have elaborated on thresholding techniques of an image from the cooccurrence matrix. These vary according to the type of exploited image [1–2]. We showed the good results of thresholding by cooccurrence matrix on images with a defect. We develop in this paper a two-part study. Firstly, after a brief account of the matrix exploitation, we show the limits of this method in bearing simulated images. We then describe the chosen approach to extend the method to the volume thresholding problem. We give results from a data volume obtained on austenitic steel testing.

Volume

15A

Chapter

Chapter 3: Signal Processing and Image Analysis

Section

Image Analysis

Pages

875-882

DOI

10.1007/978-1-4613-0383-1_114

Language

en

File Format

application/pdf

Share

COinS
 
Jan 1st, 12:00 AM

Development of a New Method for the Inspection of a 3D Volume of Ultrasonic Data

Seattle, WA

NDT technologies by ultrasounds are in constant progress. The advance is considerable for sensors and acquisition systems. The means for data processing follow this evolution. In order to meet the demand for processing very high volumes of data, we propose in this paper a method of volume automatic thresholding. The application concerns the testing of austenitic steel or else, for the nuclear industry. The thresholding techniques from the image histogram are inadequate as the ultrasonic image histogram is unimodal. The study of the image with the cooccurrence matrix which is a two dimensionnal histogram allows to clearly show the noise-defect transition. Several authors have elaborated on thresholding techniques of an image from the cooccurrence matrix. These vary according to the type of exploited image [1–2]. We showed the good results of thresholding by cooccurrence matrix on images with a defect. We develop in this paper a two-part study. Firstly, after a brief account of the matrix exploitation, we show the limits of this method in bearing simulated images. We then describe the chosen approach to extend the method to the volume thresholding problem. We give results from a data volume obtained on austenitic steel testing.