Method for shaping of improved radiation image
National Taras Shevchenko University of Kyiv, UKRAINE
- International Symposium on NDT Contribution to|
the Infrastructure Safety Systems, 1999 NOV 22-26 Torres,
published by UFSM, Santa Maria, RS, Brazil
Imaging inspection technique such as radiography, tomography and others are used in the non-destructive examination (NDE) of industrial equipment components in order to locate any cavities, inclusions, lack of fusions and so on that may have been formed during the manufacturing or operation process . Reliable detection of defect indications in radiographic images is a topic of up-to-date research in this area, including computer image processing . The most difficult problem in the inspection cycle is the accurate detection of flaws in a given radiographic image. The possibility to detect flaw at the background caused by the specimen itself to be investigated is being determined by difference in contract. One of the main methods to increase the contrast of image is to use more powerful flux or bigger exposure time. However this approach is not always acceptable and has some shortcomings. Another method to enhance the delectability of defects is an application of filtering to the image having been formed to improve the contract but taking into account false defects revealed.
The method proposed is based at statistical properties of the Poisson flux caused by ionizing radiation detection. It works at the earliest stage of image shaping when the statistical properties of incident flux have not been lost yet and includes the analysis and processing of image forming flux taking into account local properties of it in the vicinity of each event and, as a consequence, irregularities in flux density. Hence, the existing gradient in flux density between the different sections of the same image can be magnified. Due to density of image forming flux depends on the specimen internal structure one may get to improve the contract and detection of defects. From the practical point of view we are allowed to obtain the pictures of defects with enhanced contrast in comparison to images being formed without the analysis and modification of original flux.
Fig 1: Schematic diagram of image forming.
In classical schemes the flux y (x,y,z) comes to imagine surface where it is being integrated and shaped.
In the frames of the method we propose to process the flux y (x,y,z) before it will form the image. The processing itself is based at analysis of separate flux sections with respect to their own statistical properties and taking into account the correlation between different sections. Such the approach allows to change the level of detailisation and, subsequently, to adjust the sensitivity to discontinuities of small sizes as well as the contrasting amplification to big flaws. Moreover, the method proposed can be used either for image contrasting for low intensity flux by using the space processing of results or for getting of contrast image by means of flux processing in time scale.
To demonstrate the advantages of the method proposed we conducted modeling experiments to determine it's quantitative and qualitative characteristics. The model for flux simulation and subsequent image shaping has been developed and verified. It was assumed that incident flux is the Poisson one and during the exposure time the flux is constant.
The model developed realizes the diagram at Fig.1 to shaping the image and gives the possibility to include the program unit to process the flux y (x,y,z).To be able to estimate the results obtained, the contrasting coefficient K was proposed to be as a measure of contrast enhancement:
1. For the first set of experiments the procedure of image shaping was modeled. The "specimen" had the notches of different areas machined in the surface. The original "specimen" image is shown at Fig.2 a). The result of classical shaping image is presented at Fig.2 b). Fig.2 c) shows the result of implemented procedure. Also the contrasting coefficients K are given for both procedures.
|where: B0 - brightness of given part of image;
|Bb - background brightness caused by the specimen itself.
|Fig 2: Modeling results for exposure the specimen having notches of different areas machined in the specimen surface.
|Methods for image
|Table 2: Values of contracting coefficients both for "classical" and proposed method of image shaping.|
2. To estimate the method capabilities the image has been modeled for the specimen having notches of different depths, but the same area and located athwart to incident flux. Fig. 3 a) presents original image of the specimen. Fig 3 b) shows the picture, which is typical, i.e., obtained in classical way. Fig. 3 c) demonstrates the image processed with the method proposed. Table 1 contains the contrasting coefficients for different image sections. Analysis of results confirms that the method proposed allows to obtain the pictures of specimen with different density at the same image having kept the linearity of contrasting coefficients ratios for different image sections.
Fig 3: Modeling results for exposure of the specimen, having notches of different depths.
The method proposed to shaping the radiation images has the distinction from the well- known procedure. Being based at statistical properties of radiation fluxes it has some advantages to improve the quality of radiation images before they will be finally formed. One can say also that the "classical" result can be easily obtained from the method described above under some constraints. One of the most important features is additional character of presented approach, which does not influence well-known existing methods.
The results presented do not cover all the available advantages because some additional parameters can be helpful in procedure to enhance the image quality and further investigations and developments can be considered as perspective.
/DB: Article /SO: NDTISS /AU: Kadenko_I /AU: Galushka_A /AU: Yermolenko_R /CN: UA /CT: RT /CT: film /ED: 2000-05
- Nondestructive Testing Handbook. Second Edition. Volume 3. Radiography & Radiation Testing. ASNT,1985, 501 p.
- Flaw Detection in Radiographic Images by Structure-Adaptive Binary Segmentation. R. M. Palenichka, A.V. Alekseichuk, U. Zscherpel // Computerized Tomography for Industrial Applications and Image Processing in Radiography. March 15-17, 1999, Berlin, Germany DGZfP Proceedings BB67-CD, Poster 12.
- J.W. Muller: "Some remarks on the Galushka method", Rapport BIPM -93/2, Bureau International des Poids et Mesures, (1993).