RU2545503C1 - Method of determination of geometrical displacement of sensors in flat panel detector of x-ray image - Google Patents

Abstract

FIELD: instrumentation.

SUBSTANCE: scope of application: to determine geometrical displacement of sensors in flat panel detector of X-ray image. The essence of the invention is that on the work surface of the detector the test object is located, it contains at least two objects "sharp edge" corresponding to position of the process gap between the specified sensors; X-ray radiation flow is directed on the test object, its X-ray images is obtained, on the image pixels are identified that correspond to image of the sharp edge of each object "sharp edge", using them the geometrical displacement of sensors is determined on condition of objective functional minimum with limitations for the specified displacements; at that limitations include the linear limitation corresponding to the geometrical displacements of the sensors located with each other along horizontal or vertical line, and no-linear limitations corresponding to the geometrical displacements of sensors located with each other along diagonal.

EFFECT: expansion of range of technical means determining geometrical displacements of the sensors in the flat plate detector of X-ray radiation and possibility of sensor displacements determination with high accuracy.

4 cl, 11 dwg

Description

The invention relates to methods for processing digital images, in particular to a method for determining geometric displacements of sensors in a flat-panel x-ray image detector from an image of a test object.

Currently, various manufacturers of medical equipment are developing flat-panel x-ray detectors with a working field up to several tens of centimeters. Some of these detectors consist of several sensors rigidly mounted on a common substrate. For example, in [US patent No. 6895077, publ. May 17, 2005] an X-ray apparatus is described in which a detector of four (two by two) or nine (three by three) CCD sensors can be used as possible options. In the technical solution [US patent No. 7663115, publ. 02.16.2010] a detector consisting of six CMOS sensors with a working field of 20 × 30 cm is described. On the x-ray image obtained using such a composite detector, in the area corresponding to the joints of the sensors, there are various kinds of artifacts that arise for the following reasons. Firstly, the sensors differ in their photosensitive characteristics. Secondly, in an ideal detector there should be no gaps between the sensors and each column (row) of the sensor should go to the corresponding column (row) of the neighboring sensor. It is clear that in real detectors the sensors will always be geometrically offset relative to their ideal position, which also affects the image quality.

Both of these reasons lead to noticeable artifacts in the image and to the need for their correction. To correct such images, it is also important to understand the nature of such artifacts and to have additional dimensions numerically characterizing these artifacts.

Among the image correction methods, for example, a method for correcting boundary artifacts in an X-ray image is known [US patent No. 8073191, publ. 12/06/2011], based on the application of the multidimensional hypothesis of the hidden Markov model. The description of the technical solution indicates that the width of the artifact zone can reach several pixels, but the main focus of the method is on the correction of the artifacts themselves, as such.

The flat panel detector is a one-piece device and when assembled it does not allow direct measurement of the displacements between the sensors. Therefore, two methods for determining geometric displacements are possible. The first method is that the displacement of the sensors is determined by direct measurement, using measuring equipment, at the detector assembly stage. For example, a Galileo AV350 optical microscope [Galileo AV350 Multi-Sensor Vision System, the L.S. Starrett Company] allows distance measurements to within a few microns. The second method is to determine the displacements from the x-ray image of the test object.

The disadvantage of direct measurement of displacements is that, firstly, due to mechanical stresses, the location of the sensors in the assembled detector may differ from the values measured by the detector in disassembled form. Secondly, if it is necessary to carry out such measurements for an already assembled detector, then the detector should be disassembled only in a room specially equipped for this. Both of these reasons practically exclude the possibility of determining the displacements of sensors outside their production, for example, in a medical clinic.

In similar situations, when it is undesirable to disassemble the device, various indirect methods are often used. For example, a method is known for determining the geometric displacements of sensors in a scanner from an image of a test object [US Pat. No. 6,600,568, publ. 07/29/2003]. In this method, a test object is scanned with an image of a specific image, the image sections corresponding to different sensors are selected on the scan, and the displacements of the sensors are calculated from their displacements.

The claimed technical solution considers a method for determining the geometric displacements of sensors in a flat-panel x-ray detector from the image of the test object. The possibility of determining the geometric displacements of sensors from the image of a test object, which includes a substrate of X-ray transparent material and objects with a “sharp edge”, has been experimentally shown. In the areas of interest of the X-ray image of the test object, the pixels corresponding to the sharp edge of each object “sharp edge” are identified, the data for calculations are generated and the geometric displacements of the sensors are determined from the condition of the minimum of the target functional with restrictions on the indicated displacements.

The author is not known from the prior art a method for determining the displacement of the sensors, similar to the claimed.

The technical problem to be solved by the claimed invention is aimed at expanding the arsenal of means for determining geometric displacements of sensors, and more specifically, in creating a new method for determining geometric displacements of sensors from the image of a test object, which allows with sufficient accuracy to determine the displacement of sensors in a flat-panel x-ray detector .

The technical result is the expansion of the arsenal of technical means for determining the geometric displacements of sensors in a flat-panel X-ray detector and the ability to determine the displacement of sensors with high accuracy.

The specified technical result is achieved in a method for determining the geometric displacements of sensors in a flat-panel x-ray image detector containing at least two sensors mounted on a mounting panel to provide a technological gap between the sensors at the junction of the sensors, namely, that a test object is placed on the working surface of the detector comprising at least two objects "sharp edge" corresponding to the position of the technological gap between these sensors. The flow of x-ray radiation is directed to the test object and get its x-ray image. The resulting image identifies the pixels corresponding to the image of the sharp edge of each object "sharp edge", which determine the geometric displacement of the sensors from the minimum condition of the target functional with linear and nonlinear restrictions on these displacements. In this case, linear constraints correspond to geometric displacements of sensors located horizontally or vertically next to each other, and nonlinear constraints correspond to geometric displacements of sensors located diagonally next to each other.

To identify the pixels corresponding to the sharp-edge image, the image gradient modulus is calculated and pixels with a gradient modulus above a predetermined threshold value are identified, data is generated from the coordinates of such pixels and weight factors, and the pixel gradient module is used as weight factors.

The least-squares method is used as the objective functional.

It is advisable to perform the test object in the form of a substrate of X-ray transparent material, on the surface of which a marking is applied in the form of straight sections corresponding to the joints of the sensors. The number of joints is greater than or equal to one, depending on the number of sensors. At least two “sharp edge” objects are placed on each of these segments so that the rectilinear sharp edges of adjacent objects are perpendicular to each other, and the angle between the sharp edge of each object and the corresponding segment is mainly forty-five degrees, and the sharp edges of the objects divide the aforementioned a segment into essentially equal parts.

The specified set of essential features allows to achieve a technical result, which consists in determining the geometric displacements of the sensors with the necessary accuracy.

The implementation of the method for determining the geometric displacements of the sensors of a flat panel detector is illustrated by the following drawings.

Figure 1 shows a device for implementing the method, where the positions indicated:

1 - source of x-ray radiation;

2 - x-ray flux;

3 - X-ray detector;

4 - test object.

Figure 2 shows a photograph of sensors mounted on a common substrate. The photo shows that the sensors do not fit tightly to each other and there are gaps between them. Moreover, the displacements of the sensors relative to a given position lead to the fact that the sensors are displaced horizontally, there can also be angular displacements of the sensors, which leads to the formation of technological gaps between the sensors, increased or decreased compared to the set and uneven in width.

Figure 3 presents an enlarged part of the X-ray image of the worlds at the junction of the sensors of the detector. The oval marks the portion of the image in which artifacts in the area of the joints of the sensors are most noticeable.

Figure 4 shows a schematic representation of the test object 4, where the positions indicated:

I and II - image areas corresponding to the sensors of the detector;

5 - substrate;

6 - a straight line segment corresponding to the joints of the sensors of the detector;

7-8 - objects "sharp edge" and the corresponding zones of interest.

This test object is designed to determine the displacements of the sensors of a detector consisting of two sensors.

Figure 5 shows a schematic representation of the test object 4, where the positions indicated:

I-IV - image areas corresponding to the sensors of the detector;

5 - substrate;

6, 15 - straight line segments corresponding to the joints of the detector sensors;

7-14 - objects "sharp edge" and the corresponding zones of interest.

This test object is designed to determine the displacements of the sensors of the detector, consisting of four sensors (two to two sensors).

Figure 6 shows the module gradient image of the object "sharp edge".

Figure 7 presents the plot of the x-ray image of the test object with the joints of adjacent sensors. Dots mark the pixels by which the displacements of the sensors are calculated. The vertical and horizontal axes show the numbering of pixels.

On Fig. 8, explanations are given for identifying a line by a given set of points. Dots indicate the data set (x, y), on which a straight line with parameters (p, θ) is constructed.

Figure 9 shows a graph of the transfer function of the modulation of the detector, the horizontal values of spatial frequencies in mm ^{-1 are} plotted. The vertical axis represents the values of the modulation transfer function.

Figure 10 presents a histogram of the absolute errors of the displacements of the sensors along the x axis. The horizontal axis represents the absolute error values in pixels, and the vertical axis represents the corresponding probability values in percent.

Figure 11 presents a histogram of the absolute errors of the displacements of the sensors along the y axis. The horizontal axis represents the absolute error values in pixels, and the vertical axis represents the corresponding probability values in percent.

An x-ray image is obtained using the device shown in figure 1. The device contains an x-ray source 1. The x-ray stream 2 is directed to the working side of the x-ray image detector 3, on which the test object 4 is placed. The detector 3 contains a scintillation screen (not shown), which is optically connected with the active region of the detector. A scintillation screen converts X-ray radiation 2 into visible light, the sensors of the detector register it in the form of a digital image. According to the claimed method, a test object 4 is placed on the working surface of the detector 3, containing at least two sensors mounted on a mounting panel, providing a technological gap between the sensors at the junction of the sensors (Fig. 4). The x-ray flux is directed to the working side of the detector and an x-ray image of the test object is obtained.

We describe a method for determining the geometric displacements of the sensors of a flat-panel detector using an X-ray image of a test object.

On the x-ray image, the image of the sharp edge should be approximated with sufficient accuracy by a straight line segment. The essence of the method is to perform the following steps:

1) For each zone of interest, a data set is formed from the coordinates of the pixels and weight factors that correspond to the image of the sharp edge. As weighting factors, the gradient module of the corresponding pixel is used.

2) Such coordinate transformations are determined, i.e. sensor displacements so that the points corresponding to the same sharp edge lie on a straight line with the least error. As an error, i.e. target functional, use the sum of the weighted squares of the residuals.

Let us describe a method of generating data for each zone of interest, for example, positions 7-14 marked in Fig. 5 for a detector of 2 × 2 sensors. In a standard way [R. Gonzalez, R. Woods, S. Eddins, Digital image processing in MATLAB, Technosphere, 2006., p.401] calculate the image gradient modulus using a one-dimensional linear filter of radius r

Each pixel with coordinates (x _i , y _i ) is assigned a weight ω _I equal to the gradient modulus. In subsequent calculations, we will use only those pixels whose weights are higher than a given threshold value k × ω _max of the maximum pixel weight value ω _max in the corresponding zone of interest. The constant k and the parameters of the linear filter (r, σ) are selected in the course of numerical experiments. Figure 6 shows a part of the image of the gradient module. 7, pixels are marked with dots whose weights are above a predetermined threshold value.

We describe a method for identifying a straight line on a plane according to the data (x _i , y _i , ω _i ), where (x _i , γ _i ) are the coordinates and ω _i are the weights of the pixels. We write the equation of the line (p, θ) in the following form:

p + x × cosθ + y × sinθ = 0

The parameters of the straight line (θ, p) are determined from the minimum condition for the functional

which is the sum of the weighted average squares of the distances from each pixel to the line (θ, p). The same functionality can be represented in matrix form

here τ = (cosθ, sinθ) and X _i = (x _i , y _i ) ^T. The values of the parameters θ and p, delivering a minimum to the functional E (θ, p), are determined by the expressions

, $$\overline{y}=\frac{{\displaystyle \sum {\omega }_i\times y_i}}{{\displaystyle \sum {\omega }_i}}$$

и определяются из условия равенства нулю первых производных функционала E (θ, p). На фиг.8 приведены пояснения к идентификации прямой по заданному набору точек.Where $$\overline{x}=\frac{{\displaystyle \sum {\omega }_i\times x_i}}{{\displaystyle \sum {\omega }_i}}$$

, $$\overline{y}=\frac{{\displaystyle \sum {\omega }_i\times y_i}}{{\displaystyle \sum {\omega }_i}}$$

and are determined from the condition that the first derivatives of the functional E (θ, p) vanish. On Fig. 8, explanations are given for identifying a line by a given set of points.

и $${\omega }_i^{R,S}$$

обозначают координаты пикселей и веса, принадлежащие зоне интереса R и сенсору S. Введем глобальную систему координат, связанную с левым верхним углом сенсора I, в координатах которой будем проводить все вычисления. Запишем преобразование координат на плоскости (поворот и смещение) в видеLet's move on to the description of the next step in determining sensor offsets. For simplicity, we first describe a method for determining the offsets for a 1 × 2 type detector (the test object is shown in FIG. 4). Let be $$(x_i^{R,S},y_i^{R,S})$$

and $${\omega }_i^{R,S}$$

denote the coordinates of pixels and weights belonging to the zone of interest R and the sensor S. We introduce the global coordinate system associated with the upper left corner of the sensor I, in the coordinates of which we will carry out all the calculations. We write the coordinate transformation on the plane (rotation and displacement) in the form

here the matrix O and the vector D determine the coordinate transformation (rotation and displacement) on the plane

We write the target functional in the form

, S - нумерация сенсоров S=I, II, R - нумерация зон интереса R=7,8. Параметры прямых (p₇, θ₇) и (p₈, θ₈), соответствующие острым краям объектов 7 и 8, поворот O_II и смещение D_II второго сенсора определим из условия минимума целевого функционала E_I,II. При решении оптимизационной задачи следует исключить те решения, которые приводят к пересечению сенсоров, поэтому на величины O_II и D_II следуетhere τ _R = (cosθ _R , sinθ _R ) and $$X_i^{R,S}={(x_i^{R,S},y_i^{R,S})}^T$$

, S - numbering of sensors S = I, II, R - numbering of zones of interest R = 7.8. The parameters of the straight lines (p ₇ , θ ₇ ) and (p ₈ , θ ₈ ) corresponding to the sharp edges of objects 7 and 8, the O _II rotation and the D _II offset of the second sensor are determined from the minimum condition for the objective functional E _{I, II} . When solving the optimization problem, those solutions that lead to the intersection of the sensors should be excluded; therefore, the values of O _II and D _II should be

impose additional restrictions. Since the angle of rotation of the sensors in a real detector is quite small, we will consider them equal to zero. Then the boundary conditions are written especially simply and get the following optimization problem with one linear constraint for the x component of the vector D _II :

Let us turn to the description of the determination of sensor displacements for a detector with an arbitrary number of sensors. Let N mean the number of sensors vertically and M the number of sensors horizontally. To number the sensors in the detector, we will use the multi-index (n, m), where n is the sensor number vertically (from top to bottom) and m is the sensor number horizontally (from left to right).

To determine the geometry of the detector with an arbitrary number of sensors, minimize the target functional of the following form:

соответствуют стыкам между сенсорами (n, m) и (n, m+1), а слагаемые $$E_{n,m}^y$$

соответствуют стыкам между сенсорами (n, m) и (n+1, m). Слагаемые $$E_{n,m}^x$$

и $$E_{n,m}^y$$

можно представить аналогично примеру для детектора типа 1×2.Terms $$E_{n,m}^x$$

correspond to the joints between the sensors (n, m) and (n, m + 1), and the terms $$E_{n,m}^y$$

correspond to the joints between the sensors (n, m) and (n + 1, m). Terms $$E_{n,m}^x$$

and $$E_{n,m}^y$$

can be represented analogously to the example for a 1 × 2 type detector.

To the boundary conditions, together with vertical and horizontal restrictions, we should add restrictions for sensors that are located diagonally with respect to each other, such as sensors (1, 1) and (2, 2). As a result, we obtain the following optimization problem with linear and nonlinear constraints:

E → min

The first two conditions dx _1,1 = 0 and dy _1,1 = 0 appear due to the fact that the upper left sensor is selected as the coordinate system with respect to which all calculations are performed. The third group of conditions dx _{n, m + 1} -dx _{n, m} ≥0 is the essence of the constraint for displacements of sensors located horizontally next to each other. The fourth group of conditions dx _{n + 1, m} -dx _{n, m} ≥0 is the essence of the constraint for displacements of sensors located vertically next to each other. The fifth and sixth groups are the essence of the restriction for sensors that are located on the diagonal next to each other. The fifth group of conditions is the essence of the constraint for sensor offsets (n + 1, m + 1) with respect to the sensor (n, m). The sixth group of conditions is the essence of the constraint for sensor offsets (n + 1, m) with respect to the sensor (n, m + 1). In the inventive method, we used the function f (x, y) of the following form:

f (x, y} = 1-2 × logsig (-gain × (x-offset)) × logsig (-gain × (y-offset))

here logsig (x) = 1 / (1 + exp (-x)) is the sigmoid function. For gain = 100 and offset = 0.01, the function f (x, y) has the following obvious properties:

и $$d{y}_{n,m}^0=n-1$$

. Выбранные значения соответствуют тому, что величина зазора между сенсорами составляет один пиксель.These properties are guaranteed to exclude the intersection of sensors that lie on the diagonal with respect to each other. To solve this problem, standard gradient methods of numerical optimization with nonlinear constraints are used. As the initial conditions for sensor displacements in solving the problem, we used the following values: $$d{x}_{n,\mathrm{<figure-callout\; id="0"\; label="m"\; filenames="00000040.png,00000041.png"\; state="\{\{state\}\}">m</figure-callout>}}^0=m-\mathrm{one}$$

and $$d{y}_{n,\mathrm{<figure-callout\; id="0"\; label="m"\; filenames="00000040.png,00000041.png"\; state="\{\{state\}\}">m</figure-callout>}}^0=n-\mathrm{one}$$

. The selected values correspond to the fact that the gap between the sensors is one pixel.

To determine the displacements of the sensors in a flat-panel detector consisting of four (two by two) sensors, the test object 4 (Fig. 5) is placed on the working surface of the detector 3. Fig. 5 is a schematic representation of the test object, where positions I-IV areas corresponding to detector sensors are indicated, and positions 7-14 correspond to “sharp edge” objects. Frames mark the zones of interest of the image, which are used to calculate the displacements of the sensors. The test object is made in the form of a substrate 5 made of X-ray transparent material, for example, organic glass, the size of which corresponds to the size of a particular detector 3. On the substrate 5, straight sections 6 and 15 are applied. The straight lines 6 and 15 on the test object are essentially ideal sensor joints when the sensors would be placed without the necessary technological gaps in the absence of displacements of the sensors relative to each other. In real detectors during assembly, they provide technological gaps between the sensors to prevent damage to the sensors as a result of possible temperature expansions. In addition, due to inaccuracies in the assembly, the sensors are geometrically displaced relative to a given position. Rectilinear segments 6 and 15 are oriented mutually perpendicular and are located on the test object over a given area of technological gaps. On these segments, the “sharp edge” objects 7-14 are placed and fixed, ensuring compliance with the position of these objects in the technological gap area. The “sharp edge” object is a tungsten plate with a rectilinear sharp edge. The plate has dimensions of 20 mm by 10 mm (with a detector pixel size of 50 μm) and a thickness of 1 mm. A plate made in this way is used, for example, in a test object to evaluate the transmission function of the modulation of X-ray image receivers using the “sharp edge” method (GOST R IEC 62220-1-2006). Tungsten plates are placed on the straight sections 6 and 15 of the substrate 5. The best option for placing the plates is when the sharp edges of adjacent plates are perpendicular to each other, the angle between the sharp edge of each plate and the corresponding straight line segment is essentially forty-five degrees, the sharp edges of the plates divide the specified straight line segment into equal parts.

To determine the geometry of the entire detector as a whole, minimize the objective functional of the following form (N = 2 and M = 2):

определяет положение второго сенсора по первому сенсору (зоны интереса 7 и 8), $$E_{\mathrm{1,1}}^y$$

- третьего по первому (зоны интереса 11 и 12), $$E_{\mathrm{2,1}}^x$$

- четвертого по третьему (зоны интереса 9 и 10) и $$E_{\mathrm{2,1}}^y$$

- четвертого по второму (зоны интереса 13 и 14). Нумерация зон интереса 7-14 и нумерация сенсоров I-IV приведены на фиг.5. В результате получают следующую задачу с линейными и нелинейными ограничениями:Term $$E_{}^{}$$$${}_{\mathrm{1,1}}^x$$

determines the position of the second sensor by the first sensor (zones of interest 7 and 8), $${\mathrm{<figure-callout\; id="1"\; label="E"\; filenames="00000034.png,00000040.png"\; state="\{\{state\}\}">E</figure-callout>}}_{\mathrm{1,1}}^y$$

- third in the first (zones of interest 11 and 12), $$E_{2.1}^x$$

- fourth to third (zones of interest 9 and 10) and $$E_{2.1}^y$$

- fourth in the second (zones of interest 13 and 14). The numbering of zones of interest 7-14 and the numbering of sensors I-IV are shown in Fig.5. The result is the following problem with linear and nonlinear constraints:

E → min

The first two conditions dx _1,1 = 0 and dy _1,1 = 0 appear due to the fact that the upper left sensor is selected as the coordinate system. The third condition is the restriction for the displacements of the second sensor with respect to the first sensor, the fourth for the displacements of the third sensor with respect to the first sensor, the fifth and sixth for the displacements of the fourth sensor with respect to the second and third sensors. The seventh condition is a restriction for the offsets of the fourth sensor with respect to the first sensor, and the eighth condition is a restriction for the offsets of the third sensor with respect to the second sensor. To solve this problem using standard gradient methods of numerical optimization with restrictions.

As noted above, an assembled flat panel detector does not allow direct measurement of sensor offsets. For this reason, the performance of the proposed method was tested on model images. In numerical experiments, 16-bit X-ray images of a test object with a previously known sensor displacement with the following image characteristics were simulated:

1) the signal and noise levels in the air image area are 30,000 and 50 units, respectively;

2) the signal and noise level in the image area of the tungsten plate is 650 and 15 units, respectively;

3) the transmission function of the modulation of model images corresponds to the values measured from the real image, and is shown in Fig.9. The methodology for measuring the modulation transfer function complies with GOST R IEC 62220-1-2006.

генерировались генератором случайных чисел с однородным распределением в диапазоне ±2 пиксела. После этого заявляемым способом по модельным изображениям вычислялись смещения сенсоров (dx_n,m, dy_n,m), которые сравнивались с исходными значениями смещений $$(d{\tilde{x}}_{n,m},d{\tilde{y}}_{n,m})$$

. Численные эксперименты показали, что заявляемый способ позволяет определять смещения сенсоров в плоскопанельном детекторе с абсолютной ошибкой в пределах 0,2 пикселя. На фиг.10 и 11 представлены гистограммы абсолютных ошибок $${\epsilon }_x=d{\tilde{x}}_{n,m},d{x}_{n,m}$$

для смещений по оси x и $${\epsilon }_y=d{\tilde{y}}_{n,m},d{y}_{n,m}$$

, для смещений по оси y.The noise superimposed on the model image corresponds to white noise with a normal distribution. The indicated values correspond to the real x-ray image of the test object. Sensor Offsets $$(d{\tilde{x}}_{n,m},d{\tilde{y}}_{n,m})$$

generated by a random number generator with a uniform distribution in the range of ± 2 pixels. After that, by the claimed method, based on model images, the displacements of the sensors (dx _{n, m} , dy _{n, m} ) were calculated, which were compared with the initial values of the displacements $$(d{\tilde{x}}_{n,m},d{\tilde{y}}_{n,m})$$

. Numerical experiments showed that the inventive method allows to determine the displacement of the sensors in a flat panel detector with an absolute error within 0.2 pixels. Figures 10 and 11 show histograms of absolute errors. $${\epsilon }_x=d{\tilde{x}}_{n,m},d{x}_{n,m}$$

for offsets along the x axis and $${\epsilon }_y=d{\tilde{y}}_{n,m},d{y}_{n,m}$$

, for y-axis offsets.

Using the described method for determining the geometric displacement of sensors in a flat-panel x-ray detector from the image of the test object allows you to simply, efficiently and with high accuracy to evaluate the geometric displacement of the sensors without disassembling the detector. The proposed method expands the arsenal of technical means for a specific purpose.

In the above description of the invention, described in an independent claim, the possibility of its implementation using the above-described and known means and methods is shown. Therefore, the claimed method meets the condition of industrial applicability.

The proposed technical solution is disclosed in the description with possible examples of its implementation, which should be considered as illustrations of the method, but not as its limitation. Based on this description, those skilled in the art may suggest other variations within the scope of the appended claims.

Claims (4)

1. The method of determining the geometric displacements of the sensors in a flat-panel x-ray image detector containing at least two sensors mounted on a mounting panel with a technological gap between the sensors at the junction of the sensors, which consists in placing a test object on the working surface of the detector, including at least two “sharp edge” objects corresponding to the position of the technological gap between these sensors, the x-ray radiation stream is sent to the test object, I get its x-ray image on the resulting image identifies the pixels corresponding to the image of the sharp edge of each object "sharp edge", which determine the geometric displacement of the sensors from the condition of the minimum of the target functional with restrictions on these displacements, and the restrictions include linear constraints corresponding to the geometric displacements of the sensors located next to each other horizontally or vertically, and nonlinear constraints corresponding to the geometric displacements of the sensors diagonally adjacent to each other. 2. The method according to claim 1, characterized in that to identify the pixels corresponding to the image of the sharp edge, calculate the gradient module of the image and select pixels with a gradient module above a predetermined threshold value, generate data from the coordinates of such pixels and weight factors, as weight factors use a pixel gradient modulus. 3. The method according to claim 1, characterized in that the least squares method is used as the target functional. 4. The method according to claim 1, characterized in that the test object is a substrate of X-ray transparent material, on the surface of which markings are made in the form of straight sections corresponding to the joints of adjacent sensors, the number of joints being greater than or equal to one depending on the number of sensors, at least two “sharp edge” objects are placed on each of these segments so that the rectilinear sharp edges of adjacent objects are perpendicular to each other, and the angle between the sharp edge of each object and the corresponding segment nd is preferably forty-five degrees, with the sharp edges of objects are divided into equal length of said substantially part.

Images