JPS6325538A - Angle limiting image forming method - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION BACKGROUND OF THE INVENTION This invention is useful when the range of angles covered is limited.
TECHNICAL FIELD The present invention relates to imaging objects using computed tomography equipment, and in particular to methods for incorporating a priori information of the dominant components of an object into angle-limited reconstruction procedures.

X-ray CT methods are used for non-destructive inspection of industrial products and equipment. Standard CT reconstruction methods calculate the density distribution within the cross-section of the object being examined from the 7711 J constant value of the cross-section attenuation at every angle. However, in the situation where
Objects can only be scanned within a limited angular range. For example, one dimension of the object may be very long, so measurements at small angles to the longitudinal direction are too strongly attenuated to be useful. Furthermore, the measurement of the angular range may be obstructed by other objects. Images constructed from such limited data typically have artifacts.

The angle-limited replay method was developed to eliminate artifacts caused by missing information. One reference is a paper published by the inventor and V. Berret Mendez in Journal of the Optical Society of America 71 (May 1981 issue), pages 582-592, “Angle-limited ``Tomographic Imaging Using Input''. Typically,
Angle-limited image reconstruction schemes utilize available a priori information of the object to compensate for missing scan information. The most readily available a priori information is the outer boundaries of the object and the known upper and lower limits of the density values. The more precisely this a priori information can be determined, the better the quality of the reconstructed image will be. Still, images reconstructed in this manner are typically not as good as images reconstructed from complete angular scan information.

A method to further improve images reconstructed from angle-limited information utilizes multi-energy x-ray scanning as described in the inventor's US Pat. No. 4,506.327. This method is
This is for complex objects composed of a small number of substances. Industrial products and equipment usually meet this requirement. An object is scanned several times with X-rays of different energies. By appropriately combining the scan data, different parts inside the object can be reconstructed individually, resulting in much better image quality. Of course, this method cannot be used if it is impractical to perform multiple energy x-ray scans.

SUMMARY OF THE INVENTION Angularly restricted imaging of an object takes other a priori information about the object other than the ordinary a priori information, such as outer boundaries and upper and lower limits of density values. improved by adding In many cases, an object such as a metal in which a flaw is embedded has a medium occupying a large portion of its cross-sectional area, the density of which is known. The idea of this reproduction procedure is to reproduce the image of the difference between the original object and a virtual object consisting only of that medium, which has the same outer boundaries as the original object, excluding the contribution of this medium. be.

In an embodiment, a method for imaging and reconstructing flaws in such an object with limited angles is to scan the object at various angles with an X-ray or other imaging agent, and measure at each scanning angle. including generating a projected projection. Compute the projection of the virtual object mentioned earlier. The calculated projection is subtracted from the measured projection of the corresponding scan angle to generate the projection of the sword caused by the presence of the flaw. By back projecting the difference projections at several angles to create an image strip inside the boundary within which the flaw is localized, and by superimposing all the image strips, Configure the flaw-confined region(s). Using the region in which the flaw is confined and its projection, the flaw shape and density values are reconstructed by an angle-limited reconstruction algorithm.

Another aspect of the invention is that the backprojected image strip has a non-zero value, and all other image portions that are inside the boundary are assigned a zero value.

Yet another aspect is that the angle-limited regeneration procedure is iterative;
The reconstructed image is transformed back and forth between the object space by the filtered back projection and the projection space by the projection, and a priori information about the region in the object space where the flaw is confined (the range and location of this region, the density (upper and lower limits) and the projection of the difference in projection space.

The object 10 in FIG. 1 is composed of a medium m that occupies most of its area, but it has one or more flaws 11. The density of the medium ρ□ is known, and the outer boundary S of the object
is known. This is the case in industrial non-destructive testing where the object is composed of a metallic medium and there are some scratches buried inside it, and the scratches are usually voids or inclusions. , but not limited to this case. "Wound"
The word ``m'' is a nickname used in a broad sense to refer to substances that are flaw components other than the medium m inside an object.

The regeneration procedure is shown in FIGS. 2-5.

Although one flaw is shown, the method can be used to reconstruct and image several flaws in the same way. In FIG. 2, the original object A is scanned using an imaging agent, such as an X-ray, and from the detected data a measured projection pl is generated. This is done at many scan angles over the limited angular range available.

pI, 1)2. ... pn with n angles θ1. θ2.・
...θ. Let each represent the measured projection. medium m
A virtual object B is used, which is made up of An image that is the difference between the original object A and the virtual object B is reconstructed. This eliminates the contribution of medium m. To proceed with playback, the outer boundary S and the value ρ□
From the knowledge of , calculate the projection q of virtual object B. Projections Ql, Q2...Qn of virtual object B are calculated at n angles. At each scan angle, the calculated projection q, is subtracted from the measured projection p1, as shown at 12, to produce a projection d, of the difference caused by the presence of the flaw.

Several projections q1 are subtracted from the corresponding projections p1 and the difference in projections is denoted by di -1)+ -C1i. If there are multiple flaws, the difference projection will have more than one non-zero region at most scan angles.

Referring to Figure 3, the next step in the reconstruction procedure is to back project the difference projection so that the flaw is localized within it.
The purpose is to derive the image strips inside the boundary S. Each difference projection d, is back-projected at a respective scan angle θ1. D.
Let denote the portion of the backprojected image that is inside the boundary and has a non-zero value. The portion / corresponds to the region of the original object A that is occupied only by the medium m. All the flaws are localized to part Di1, ie to the strip of the image.

In hl shown in FIG. 4, all the image strips back-projected,
are superimposed to form an area that confines flaws. In this figure, the projections of the differences in five scan angles spanning less than 180° are indicated by dl to d5. The area at the intersection of the five image strips encompasses all flaws in the object. The area does not need to be a single area. It may be composed of a number of separate convex regions.

The next step is shown in FIG. Using the region where the flaw is confined and the difference projection d1, the flaw(s) 11 is regenerated by a known angle-limited regeneration algorithm. First, all other component substances or scratches are separated from the boundary S.
It is known that it is somewhere inside. After the step shown in FIG. 4, the region in which it occurs has been narrowed from S to a much smaller region. This time, since the area in which the flaw occurs is known much more precisely, the area in which the flaw is confined is used as a priori information to reproduce the limited angle of the flaw. Even better results can be obtained.

One suitable angle-limited regeneration scheme, published in the previously cited paper, is shown in FIG. This is an iterative algorithm. The reconstructed image is transformed back and forth between object space by filtered backprojection and projection space by projection, and is iteratively transformed using a priori information about the object in object space and a known projection in projection space. Corrected. The known information about the object, ie, the area in which the flaw is confined, is the range and location of this area, and the upper and lower limits of the density value. Determine the density of the object by estimation, reset pixels outside the known range of the object to zero, pixels with density above the upper limit to the upper limit, pixels with density lower than the lower limit to the lower limit Correct by

Projection of the difference between objects in a limited angular range dl to d
5 and initially set the other projections that are missing to make up the complete 180° angular range to zero. From the first set of projections, a first estimate of the density of the object is obtained, and with the a priori information about the object described earlier,
Correct this estimate. From this second estimate of the object's density, calculate the projection at the angle of auxiliary dropout. The calculated projection of the auxiliary angle is combined with the already known difference projections at the other angles to create a new estimate of the object's density and the process is repeated. An indication of the shape of the flaw(s) as well as its density is calculated. The difference between the density of the flaw and the density of the medium is actually calculated.

FIG. 7 shows an X-ray CT image of a flaw in an otherwise homogeneous object, as an example of the present invention.
The device is shown. The x-ray beam emitted by source 13 is collimated into parallel lines by collimator 14, which passes through object 15 and is detected by x-ray detector 16. The object is scanned linearly in the direction shown and then pivoted in a staggered manner to scan at other angles over the available range. The detection signal, ie the projection p, is supplied to the processing computer 17 together with the value of θ1. A priori information about the boundaries S of the object as well as the known density p□ of the medium constituting the bulk of the object are also input to the computer 17 . Given S, ρ, and the scan angle θ1, calculate at 18 the calculated projection qi (FIG. 2) of the virtual object B at the corresponding scan angle. Calculated projection Qi to measured projection 1)
By subtracting from i, the projection d of the difference at each angle θ1 is determined at 19. 20, by back projecting the difference projection dI at the corresponding angle and determining the intersection of the image strips, (Fig. 3), the area enclosing the flaw(s). The boundary between the lines is found (Figure 4). Finally, by using the projection d1 of the area D1 difference in which the flaw is confined and the value of the scanning angle θ1 for angle limited reproduction, the flaw at 21 is reproduced and displayed at 22. An indication of the shape of the flaw(s) and its density is displayed and, if desired, the boundary S of the object is displayed.

The application of X-ray CT for industrial non-destructive evaluation has recently increased. Angle-limited x-ray imaging is frequently performed in industrial inspections. The situations in which this invention is used are 18
This may be the case if measurements over 0' or the full 3600 are obstructed by other objects, or if the object is long and thin so that measurements at vertically close angles do not provide useful information.

Although preferred embodiments of the invention have been illustrated and described,
Those skilled in the art will appreciate that various modifications may be made within the scope of this invention.

[Brief explanation of the drawing]

Figure 1 is a diagram of a typical industrial object composed of a medium containing many flaws, and Figure 2 is a projection of the difference between the measured projection and the calculated projection of a virtual object without flaws. Figure 3 is a diagram showing the back projection of each difference projection, Figure 4 is a diagram showing how to construct a region that envelops all flaws in the object, Figure 5 is the angle limit A schematic diagram illustrating how a regeneration algorithm regenerates a flaw using a constructed flaw-enclosing region and difference projection; FIG. 6 is a flowchart of one angle-limited regeneration method; FIG. The range is 1. FIG. 2 is a block diagram showing an X-ray computed tomography apparatus for industrial inspection when the present invention is limited.

Claims (1)

[Claims] 1) In an angle-limited imaging method for reconstructing flaws buried inside an object using a computed tomography device, an imaging agent is used at many angles over a limited angular range. scan the object to produce projections measured at all scan angles, consisting only of a known medium that occupies most of the cross-sectional area of the original object and has the same external boundaries as the original object. subtracting the calculated projection from the projection measured at each scan angle to generate the difference projection caused by the presence of the flaw, and back projecting the difference projection. construct a flaw-confined region by determining the intersection of the image strips inside the boundaries of the object in which the flaw is localized, and use the projection of said difference and the flaw-confined region. An angle-limited imaging method comprising the step of reconstructing the flaw using an angle-limited reconstruction algorithm. 2) In the angle-limited imaging method as claimed in claim 1), the image strip has a value other than zero and all other image areas inside the boundary have a value of zero. Angle-limited imaging method. 3) The angle-limited imaging method according to claim 2), wherein the image forming agent is X-rays. 4) In the angle-limited image forming method described in claim 1), the angle-limited reconstruction algorithm is iterative, and the reconstructed image is moved between an object space by filtered back projection and a projection space by projection. An angle-limited imaging method that is transformed back and forth and iteratively corrected by a priori information about the flaw-confining region in object space and a projection of the difference in projection space. 5) In the angle-limited imaging method described in claim 4), a priori information regarding the region in which the flaw is confined is
An angle-limited imaging method that is the range and location of the region, and the upper and lower limits of the density value. 6) In a method of imaging with a computed tomography apparatus a flaw in an object containing a medium occupying a large portion of the cross-sectional area, the external boundaries of which are known and the density of which is known. An object is scanned with X-rays at many angles over a limited angular range, producing projections measured at all scanning angles, which are composed only of the medium and have the same outer surface as the original object. Calculate the projection of the virtual object with boundaries, subtract the calculated projection from the projection measured at each scan angle to create a difference projection due to the presence of the flaw, and backproject the difference projection. deriving an image strip inside the boundary in which the flaw is localized, superimposing the back-projected image strips to constitute an area in which the one or more flaws are confined; A method comprising the steps of: regenerating a flaw by an angle-limited regeneration procedure using a projection of the image and a region containing the flaw, and displaying an indication of the shape of the regenerated flaw and its density. 7. A method as claimed in claim 6) in which a strip of the back-projected image has a value other than zero and all other image parts inside the boundary have a value of zero. . 8) In the method described in claim 6), the angle-limited reproduction procedure is iterative, and the reproduced image is transformed back and forth between an object space by filtered back projection and a projection interval by projection, A method in which the flaw is iteratively corrected using a priori information about the region in object space that confines the flaw and a projection of the difference in projection space. 9) The method according to claim 8), wherein the a priori information regarding the region in which the flaw is confined is the range and location of the region, and the upper and lower limits of the density value.