RU2722620C1 - Method of obtaining and processing images formed using proton radiation - Google Patents

Abstract

FIELD: proton radiography.

SUBSTANCE: invention relates to proton radiography, particularly to methods of processing images formed with the help of proton radiation, and can be used, for example, in digital photography systems to determine the internal structure of objects or to study fast processes. Method involves obtaining digital images of a proton beam using at least two recording systems, the first of which is placed to the field of study, and the second – after, to obtain at least one image to the field of study and at least two images from different views after the field of study by passing a proton beam through the magnetooptical system, an auxiliary object placed before the first recording system, and the study area, processing the obtained images of the auxiliary object by selecting projective transformations, translating images obtained from the first recording system, to the image view from other recording systems, then, performing pixel-by-pixel division of last images on the resulted images from the first registration system, obtaining images of the field of research from different angles, reducing them to one aspect angle. Thereafter, the accuracy of bringing the images to the same angle is evaluated by using the test object in the form of two plates orthogonal to each other, which is placed in the study region, obtaining its image, at that, by turning and changing the plates inclination, their minimum thicknesses on the images are achieved, which indicates orientation of the plate plane strictly along the proton beam, then previously selected additional projective transformations are used to reduce plate images to one view, then, processing said images, constructing intensity profiles of the proton beam perpendicular to the image of plates with subsequent approximation thereof by Gauss function, such a procedure is repeated for images from all cameras of recording systems installed after the object of analysis, and when comparing coordinates of centers of all obtained intensity profiles, average or mean square error of reducing images to one angle is calculated to determine degree of reliability of comparison and analysis.

EFFECT: high accuracy of processing recorded proton images.

4 cl, 7 dwg

Description

The invention relates to the field of proton radiography, in particular to methods for processing images formed using proton radiation, and can be used, for example, in digital shooting systems to determine the internal structure of objects or to study fast processes.

When registering proton images, there are geometric distortions associated with magnetic optics, with a non-perpendicular arrangement of the scintillator and beam, with different angles, under which the proton images are captured by various registration channels. When processing registered proton images obtained by different registration channels, all images lead to the same angle for subsequent analysis. To carry out the analysis, that is, comparing the boundaries reconstructed from the experimental and calculated images, it is necessary to know how accurately the images are brought to one angle.

The challenge facing the field of technology is to obtain reliable information about the studied objects.

A known method of obtaining and processing images formed using proton radiation (D. Varentsov, O. Antonov, A. Bakhmutova, CW Barnes, A. Bogdanov, CR Danly, S. Efimov, M. Tndres, A. Fertman, AA Golubev, DHH Hoffmann, B. Lonita, A. Kantsyrev, Ya.E. Krasik, PM Lang, I. Lomonosov, FG Mariam, N. Markov, FE Merrill, VB Mintsev, D. Nikolaev, V. Panyuishkm, M. Rodionova, M Schanz, K. Schoenberg, A. Semennikov, L. Shestov, VS Skachkov, V. Turtikov, S. Udrea, O. Vasylyev, K. Weyrich, C. Wilde, A. Zubareva, Commissioning of the PRIOR proton microscope, arxiv : 1512.05644v2 [physics.ms-det] Jan 19, 2016). The method includes obtaining images of the proton beam using a registration system by passing it through a magneto-optical system (MOS) and the study area, in which the test object is first installed, and then it is replaced by the studied object and the subsequent processing of the obtained images. The test object is a copper square substrate with holes in the nodes of the orthogonal mesh, applied to an area of 9 × 9 mm. The test object is installed counter to the flow of protons. When a proton beam is passed through a test object, an image is obtained that establishes the correspondence between the dimensions of the test object in the resulting image and its actual geometric dimensions by spatial calibration (taking into account the distance between the extreme elements in the horizontal and vertical directions), which is used when processing the image of the object research.

The disadvantage of this method is that this test object cannot be set strictly perpendicular to the axis of the MOS, which does not completely eliminate the effect on the proton image of inaccuracies in the positioning of the object of study in the angle relative to this axis and reduces the accuracy of image processing.

A known method for obtaining and processing images generated using proton radiation (Physics Division Progress Report 1999-2000 Proton Radiography, DA Clarc et al, pp 156-168), including obtaining two digital images of the proton beam before passing through the study area using the first and the second registration systems and digital images of the proton beam after passing through the research area in the focus plane of the MOS using the third registration system. Each of the registration systems includes a converter that converts proton radiation into photons recorded by the CCD. The first image of the proton beam is obtained directly in front of the diffuser, the presence of which is necessary for further image processing and which is placed in the magneto-optical channel. The second image is obtained at a considerable distance from the diffuser - 6 m. Next, the obtained digital images are processed and the image of the study area is obtained by calculation. In this case, the following operations are carried out. Using the first two images, the image of the proton beam in the study area / plane of the object of study is obtained by calculation, then the pixel-by-pixel division of the third image by the calculation obtained to obtain the image of the study area is performed. This method is selected as the closest analogue.

The disadvantage of this analogue is that the use of a diffuser complicates the method due to the need to place it in a certain place, with the appropriate setting of the magneto-optical channel. In addition, the method is based on obtaining by calculation the image of the proton beam in the field of study / plane of the object of study from two images of the proton beam obtained before passing through the field of study, based on the linearity of the beam transformation.

To calculate this transformation, it is necessary to calculate the beam parameters (center coordinates, its width) on two images, which is problematic with sufficient accuracy, and this significantly complicates the method. When processing images, there is no automatic “stitching” of images of beams recorded before and after passing through the study area. In addition, in dynamic experiments, images are obtained from one angle, and to increase informational content, it is necessary to obtain images from different angles.

A known method of obtaining and processing images formed using proton radiation according to patent RU 2604723 (publ. 10.12.2016), partially eliminating the disadvantages of the above method and selected as the closest analogue. The method includes obtaining digital images of the proton beam using at least two recording systems, the first of which is placed before the study area, the second after, to obtain at least one image to the study area and at least two images from different angles after the study area by passing a proton beam through the MOS, an auxiliary object located in front of the first registration system, and the study area. Each registration system includes a converter that converts proton radiation into photons recorded by a camera (CCD). The registration system may include several cameras from different angles. The auxiliary object is made in the form of metal reference elements placed in one plane. Next, they process the obtained images of the auxiliary object by selecting projective transformations that translate the images obtained from the first registration system. To view the images from other registration systems, then, performing pixel-by-pixel division of the last images into the above images from the first registration system, the images of the study area are obtained from different angles. , reducing them to one angle, selecting additional projective transformations,

The disadvantage of the closest analogue is that in the framework of this method they do not determine the accuracy of reducing protonograms to one angle, that is, the accuracy of the “stitching” of images obtained from different angles. Meanwhile, when comparing the boundaries reconstructed from images, this accuracy must be known to determine the degree of reliability of the comparison and analysis.

The technical result of the proposed method is to increase the accuracy of processing registered proton images.

The specified technical result is achieved due to the fact that in the method of obtaining and processing images generated using proton radiation, which includes obtaining digital images of the proton beam using at least two registration systems, the first of which is placed before the study area, the second after , to obtain at least one image to the study area and at least two images from different angles after the study area by passing a proton beam through a magneto-optical system, an auxiliary object located in front of the first registration system, and the study area, processing the obtained images of the auxiliary object by selecting projective transformations that transfer the images obtained from the first registration system to the perspective of images from other registration systems, then, performing pixel-by-pixel division of the last images into reduced images from the first registration system, images of the region are obtained studies from different angles, reducing them to one angle, selecting additional projective transformations, it is new that after reducing the images obtained from different angles to one angle, they evaluate the accuracy of bringing the images to one angle, for which, using a test object in the form of two plates mounted orthogonally to each other, which is placed in the study area, an image is obtained, while rotating and changing the slope of the plates achieve their minimum thicknesses in the images, which indicates the orientation of the plane of the plates strictly along the proton beam, then additional projective projections selected previously the transformations are used to reduce the images of the plates to one angle, then, processing these images, build the intensity profiles of the proton beam perpendicular to the image of the plates with subsequent approximation by their Gaussian function, this procedure is repeated for images from all cameras of registration systems installed after the object of the study, and when comparing the coordinates of the centers of all the obtained intensity profiles, the average or standard error of reducing the images to one angle is calculated to determine the degree of reliability of the comparison and analysis.

To select additional projective transformations, the same auxiliary object placed in front of the first registration system that was used earlier can be used.

To select additional projective transformations, an auxiliary object located in the study area and representing a substrate with the same elements in the nodes of the orthogonal lattice can be used; according to the obtained image of the substrate elements, a correspondence is established between the coordinates of the elements in the image with the actual known coordinates by determining the pixel coordinates of the projective transformation, allowing you to translate the established pixel coordinates into known coordinates in the plane of the test object.

To select additional projective transformations, an auxiliary object located in the study area and representing a set of bands located orthogonally to each other can be used, the correspondence between the coordinates of the strip intersection nodes in the image with the actual known coordinates is established by determining the pixel coordinates of the projective transformation, allowing you to translate the established pixel coordinates into known coordinates in the plane of the test object.

Assessing the accuracy of bringing images to the same angle in different directions will determine the degree of reliability of comparison and analysis.

Using a test object, which is two plates mounted orthogonally to each other, allows using an image representing a thin line of each plate to obtain a clearly defined minimum when processing images on the proton beam intensity profile perpendicular to this line, which increases the accuracy of processing. In order to measure the error of combining images in different directions, it is proposed to install two metal plates perpendicular to each other.

By rotating and changing the inclination of the test object to obtain minimum plate thicknesses in the image, which indicates the orientation of the plane of the plates strictly along the proton beam, it simplifies information processing, and express analysis can be carried out even without using special programs. By changing the position of this minimum, one can judge the accuracy of image alignment in the direction perpendicular to the corresponding plate.

Establishing a correspondence between the dimensions of the test object in the resulting image and its actual geometric dimensions by determining the pixel coordinates of all elements of the substrate and selecting a projective transformation that allows you to translate the established pixel coordinates into known coordinates of the elements in the plane of the test object allows you to more accurately determine the coordinates of the elements found in the plane test object.

In FIG. 1 shows an auxiliary object, which is a metal reference objects placed in one plane. In FIG. 2 - test object "lattice", which additionally calculate the projective transformation. In FIG. 3, 4 - variants of test objects for conducting accuracy measurements. In FIG. 5 is a protonogram of the test object of FIG. 3. In FIG. 6 - intensity profile of the proton beam, perpendicular to the plates. In FIG. 7 is an intensity profile of a proton beam from images obtained from all cameras.

As an example of a specific implementation of a device that allows the implementation of the inventive method, a device that is based on the current U-70 synchrophasotron built in Protvino [News and Problems of Fundamental Physics, No. 1 (5), 2009, p. 32-42], and includes a camera for placing the object of study, a system for the formation and registration of a proton image. The formation system is an MOS of magnetic lenses and a collimator. The registration system includes three registration channels, consisting of a scintillation converter, a mirror and digital cameras. The equipment of the first registration system was installed directly in front of the zone of placement of the object of study (research area), while an auxiliary object, metal reference objects (Fig. 1), placed in one plane, was installed directly in front of the converter of the first registration system. After the zone of placement of the object of study, a MOS was placed, the magnetic quadrupole lenses of which are tuned to the calculated energy of the proton beam and provide focusing of protons from the plane of the object to the image registration plane. Two other registration systems were installed after the area of the facility. For additional measurements, a test object was used (Fig. 2), which is a set of 110 high-precision steel balls with a diameter of 9 mm mounted on an organic glass substrate in nodes of an orthogonal lattice with a strict interval (20 mm) between each other. The seats for mounting the balls are made on a CNC machine with an accuracy of ± 10 microns. In the center of the substrate, perpendicular to it, a tube is fixed. To take measurements of the accuracy of reducing the images to one angle, a test object was used (Fig. 3), which can be used as thin metal plates with a thickness of tens or hundreds of microns, for example, a razor blade. In order to measure the error of image alignment, it is proposed to install two metal plates perpendicular to each other. The selected test object is two blades that are glued orthogonally to each other and mounted on a substrate.

An example of a specific implementation of the proposed method can be a method of obtaining and processing images formed using proton radiation, including the following operations.

Set up a system for obtaining proton images of the object of study with three registration systems and MOS, providing focusing of protons from the plane of the object to the image plane. In front of the converter of the first registration system, 4 reference objects are placed (Fig. 1). Registration systems include three cameras. The formed proton beam is passed through the object of study. Three digital images of the auxiliary object to the area of the object of study and six images with the object of study recorded at the focusing site of the proton beam using the second and third registration systems after passing the beam through the study object are obtained. They process the obtained images of the auxiliary object, for which they determine the coordinates of the reference objects in the images before the installation area of the object of study and after (search for the centers of all four reference objects in the images or calculate the coordinates using mathematical methods of compensation) and select the projective transformations using which they are reduced in pairs digital images to one angle, after which the reduced images are received before the zone of placement of the object of study and after. Then, performing pixel-by-pixel division of the last images into the images from the first registration system, images of the study area are obtained from different angles, reducing them to one angle, selecting additional projective transformations.

To eliminate geometric distortions associated with different angles, under which proton images are shot by various registration systems, and to bring the received digital images to the correct angle, a test object is placed in the camera to place the object of study (Fig. 2). By adjusting the test object perpendicular to the axis of the MOS, we can almost completely eliminate the influence on the proton image of inaccuracies in the positioning of the test object in an angle relative to the axis of the MOS. The perpendicularity of the substrate of the axis of the MOS is checked by a digital image of the proton beam, which is passed through the MOS and the camera with the test object by angular movement of the test object, until the through hole of the tube in the image corresponds to the actual geometric size. The thus obtained image of the test object is used to determine the correspondence between the dimensions of the test object in the resulting images and its actual geometric dimensions. The correspondence between the dimensions of the test object during end-to-end calibration of the scaled transfer coefficient of the image in the image formation and registration path is established by determining the pixel coordinates of the centers of all steel balls and selecting a projective transformation that allows you to translate the established pixel coordinates to the known coordinates of the centers of steel balls in the plane of the test object. Having determined the projective transformation, they use it when processing images of the studied objects, for which they replace the test object with the studied object and obtain a digital image of the proton beam, passing it through the MOS and the camera with the object. Then, the projective transformation selected when using the image of the test object is used when processing images of the object of study. Thus, all proton images obtained upon registration of the object of study are reduced to one angle with a clearly known scale.

Next, assess the accuracy of bringing the images to one angle. To do this, using the test object (Fig. 3), get its image. By rotating and changing the tilt of the stand, the minimum thickness of the blades in the image is achieved, which indicates the orientation of the plane of the blades strictly along the proton beam (Fig. 5). According to the selected projective transformation, all registered images are reduced to a single angle with a clearly defined scale (for example, with a pixel size of 100 microns). Then, intensity profiles of the proton beam are built perpendicular to the blades (along the dashed lines in Fig. 3) and approximated by a Gaussian function (Fig. 6) with four parameters (a, b, σ and d):

Next, carry out such a procedure for the images from all registration channels (Fig. 7), the units of reference along the Ox axis are pixels with a size of 100 μm).

By analyzing the parameter b (Gaussian center) for all images, the average (or root-mean-square) error of reducing the images to one angle is calculated.

T.O. The claimed method allows to increase the reliability of information about the studied objects.

Claims (4)

1. A method of obtaining and processing images generated using proton radiation, including obtaining digital images of the proton beam using at least two recording systems, the first of which is placed before the study area, the second after, to obtain at least one image to the study area and at least two images from different angles after the study area by passing a proton beam through the magneto-optical system, an auxiliary object placed in front of the first registration system, and the study area, processing the obtained images of the auxiliary object by selecting projective transformations that translate images obtained from the first registration system , to the angle of images from other registration systems, then, performing pixel-by-pixel division of the last images into the above images from the first registration system, receive images of the study area from different angles, reducing them to a single angle, Paradise additional projective transformations, characterized in that after reducing the images obtained from different angles to one angle, they evaluate the accuracy of bringing the images to one angle, for which, using a test object in the form of two plates installed orthogonally to each other, which is placed in the study area, get its image, while rotating and changing the slope of the plates achieve their minimum thickness on the images, which indicates the orientation of the plane of the plates strictly along the proton beam, then the previously selected additional projective transformations are used to reduce the images of the plates to one angle, then processing these images, build the intensity profiles of the proton beam perpendicular to the image of the plates with subsequent approximation by their Gaussian function, this procedure is repeated for images from all cameras of the registration systems installed after the object of study, and when comparing the coordinates of the centers of all Radiation intensity profiles calculate the mean or root-mean-square error of reducing the images to a single angle to determine the degree of reliability of comparison and analysis. 2. The method according to p. 1, characterized in that for the selection of additional projective transformations use the same auxiliary object located in front of the first registration system that was used previously. 3. The method according to p. 1, characterized in that for the selection of additional projective transformations use an auxiliary object located in the study area and representing a substrate with the same elements in the nodes of the orthogonal lattice, the correspondence between the coordinates of the elements in the image with the actual known coordinates by determining the pixel coordinates of the projective transformation, which allows you to translate the set pixel coordinates to known coordinates in the plane of the test object. 4. The method according to p. 1, characterized in that for the selection of additional projective transformations use an auxiliary object located in the study area and representing a set of bands located orthogonal to each other, according to the received image of the bands, the correspondence between the coordinates of the nodes of the intersection of the bands in the image with the actual known coordinates by determining the pixel coordinates of the projective transformation, which allows you to translate the set pixel coordinates to known coordinates in the plane of the test object.

Images