Image Correction

By using a multi-row detector matrix scanner in X-ray inspection of cargo, local mutual parasitic displacement is determined and corrected, the image artifact problems caused by unevenness in detector movement and cargo speed are solved, and image quality and accuracy are improved .

CN112997213BActive Publication Date: 2024-11-12SMITHS DETECTION FRANCE SAS
View PDF 10 Cites 0 Cited by

Patent Information

Application Number
CN201980065314.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2018-10-01
Filing Date
2019-09-27
Publication Date
2024-11-12
Estimated Expiration
2039-09-27

AI Technical Summary

Technical Problem

During the process of checking goods with X-rays, the parasitic displacement of the detector and the unevenness of the cargo speed lead to image artifacts, which reduces image quality, and the prior art is difficult to effectively solve these problems.

Method used

By using a matrix scanner including at least two rows of detectors, images of the goods are generated and local mutual parasitic shifts are determined, total mutual parasitic shifts are calculated, corrected images are generated to eliminate artifacts, avoiding the use of expensive mechanical compensation equipment and hypothetical images deal with.

Benefits of technology

Improve image quality, realize the correction of detector motion and cargo speed inequality, improve image accuracy and clarity, and avoid unnecessary processing of the image.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112997213B_ABST
    Figure CN112997213B_ABST
Patent Text Reader

Abstract

In one embodiment, a method for inspecting cargo using X-rays is disclosed, comprising: scanning cargo with a matrix comprising at least two rows of detectors, wherein each region of the cargo irradiated by a first X-ray pulse is irradiated by at least one second X-ray pulse, and radiation corresponding to the first X-ray pulse is detected by a first row of the matrix, and radiation corresponding to the at least one second pulse is detected by at least one second row of the matrix; generating a first image of the cargo for the first row of the matrix and at least one second image of the cargo for the at least one second row of the matrix; determining, for each region of the cargo irradiated by the first X-ray pulse and by the at least one second X-ray pulse, a local mutual parasitic shift between the cargo and the matrix by determining a local mutual shift between corresponding pixels of the first image and the at least one second image; determining a total mutual parasitic shift by summing the determined local mutual shifts; and generating a corrected image of the cargo without the mutual parasitic shift based on at least one of the first image and the at least one second image and based on the determined total mutual parasitic shift.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to, but is not limited to, a method for inspecting cargo using X-rays, and also relates to, but is not limited to, a corresponding inspection system and a corresponding computer product or computer program. Background Art

[0002] Checking cargo with X-rays can be performed by displacing the cargo and a scanner including a detector relative to each other, scanning the cargo, and detecting with the scanner radiation generated by a plurality of consecutive X-ray pulses irradiating the cargo during the relative scanning displacement.

[0003] In the moving mode, displacing the cargo and the scanner relative to each other includes keeping the cargo stationary relative to the ground and displacing the scanner relative to the ground during the mutual scanning displacement.

[0004] In the pass-through or gate mode, displacing the cargo and the scanner relative to each other includes maintaining the scanner stationary relative to the ground and displacing the cargo relative to the ground during the scanning displacement.

[0005] In mobile mode, even on flat ground, the detector may have vertical and longitudinal (e.g., in the direction of mutual scanning shift) parasitic shifts (e.g., such as vertical shift, rotational shift, and / or pendulum shift), which may cause image artifacts and reduce image quality. The parasitic shifts of the detector may produce waves, so that most true straight lines may not appear as straight lines on the image generated based on the inspection.

[0006] In pass-through mode, the speed of the cargo may be irregular, which can also cause image artifacts and reduce image quality. The non-uniformity of speed can cause horizontal stretching or compression on the image. These artifacts appear as local horizontal parasitic shifts. Summary of the invention

[0007] Aspects and embodiments of the present invention are set out in the appended claims.These and other aspects and embodiments of the present invention are also described herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Embodiments of the present disclosure will now be described by way of example with reference to the accompanying drawings, in which:

[0009] Figure 1 A flow chart illustrating an example method according to the present disclosure is shown;

[0010] Figure 2 Schematically illustrates a device configured to at least partially implement Figure 1 Example systems of example methods;

[0011] Figure 3An example matrix according to the present disclosure is shown;

[0012] Figure 4 Schematically illustrates an example zone Z1 of cargo irradiated by a first X-ray pulse P1 and by at least one second X-ray pulse, namely pulses P2, P3, P4 and P5;

[0013] Figure 5 schematically illustrates an exemplary mutual parasitic vertical shift PDy superimposed on a mutual scanning shift SD;

[0014] Fig. 6A and 6B Examples of simplified images of different rows are schematically shown, where Fig. 6A An image without correction for scan shift SD is schematically shown, and Figure 6B Schematically shows an image with correction for scan shift SD;

[0015] Figure 7 Schematically shows the shift field and the estimation of the instantaneous shift field between two images from two different detection rows (e.g., row 1 and row 5);

[0016] Fig. 8A The coordinates of the acquired images are schematically shown;

[0017] Figure 8B Schematically shows Fig. 8A The coordinates of the actual position of the acquired image; and

[0018] Fig. 9 It is schematically shown how, for a given pixel (i', j'), the pixel (i, j) whose image overlaps (i', j') due to distortion is determined.

[0019] In the drawings, similar elements have the same reference numerals. DETAILED DESCRIPTION

[0020] Overview

[0021] Embodiments of the present disclosure provide a method for inspecting cargo. Inspection of cargo includes scanning cargo using a scanner including a detector matrix, the detector matrix including at least two rows of detectors. In some examples, the scanner may also include a radiation source. Scanning also includes displacing cargo and the matrix relative to each other. During scanning, an area of ​​the cargo is irradiated by a number of X-ray pulses, and the corresponding radiation is detected by a number of rows. The method also includes generating an image of the cargo for each row of detectors, and determining local mutual parasitic shifts (e.g., any local differences between the expected mutual scanning shift and the actual mutual scanning shift) between corresponding pixels of the generated image. The method also includes generating a correction image using one or more generated images without a total mutual parasitic shift corresponding to the sum of the local mutual parasitic shifts. The mutual parasitic shift may correspond to parasitic motion of the matrix, such as vertical shift, rotational shift, and / or pendulum shift when the matrix is ​​mounted on a cantilever of a mobile inspection system, or to parasitic motion of the cargo, such as speed non-uniformity of cargo passing in front of the matrix.

[0022] Embodiments of the present disclosure may achieve improvements in the quality of images generated based on inspections. Embodiments of the present disclosure may not involve expensive and cumbersome mechanical compensation equipment dedicated to suppressing (e.g., reducing) parasitic motion of a matrix (e.g., when the matrix is ​​mounted on a cantilever of an inspection system). Embodiments of the present disclosure may not involve image processing based on prior assumptions that may not be true or affect other parts of the image that do not require correction.

[0023] Detailed description of specific embodiments

[0024] Figure 1 A flow chart illustrating an example method 100 according to the present disclosure is shown. Figure 2 Figure 1 The method 100 is shown in Figure 2 An inspection system 10 for inspecting cargo 11 using X-rays is shown and may be used to at least partially perform the method 100 . Figure 3 Schematically, details of the matrix 12 of the system 10 are shown. Figure 2 In FIG. 1 , the inspection system 10 includes a cantilever 14 on which the matrix 12 can be mounted. Figure 2 and 3 In the example shown, the matrix 12 includes radiation detection lines, such as X-ray detection lines, among other conventional electrical elements.

[0025] like Figure 1 As shown and discussed in more detail below, method 100 may include:

[0026] At S1 , the cargo 11 is scanned using a matrix 12 comprising at least two rows 121 of detectors 122 ;

[0027] At S2, a first image of the goods 11 for the first row 121-1 of the matrix 12 and at least one second image of the goods for at least one second row 121-2 of the matrix 12 are generated;

[0028] At S3 , determining a local mutual parasitic displacement between the goods 11 and the matrix 12 , for example by using at least two images;

[0029] At S4, a total mutual parasitic shift is determined by summing the determined local mutual shifts of the illumination area; and

[0030] At S5 , a corrected image of the goods is generated based on the determined total mutual parasitic shifts.

[0031] like Figure 1 As shown, the method 100 may include scanning the cargo 11 using a matrix 12 including at least two rows 121 of detectors 122 at S1 .

[0032] In some examples, the scanning at S1 may include:

[0033] The goods 11 and the matrix 12 are mutually shifted by a mutual scanning shift SD (for example Figure 4 shown);

[0034] as well as

[0035] During a mutual scanning shift SD, the radiation 13 generated by a plurality of consecutive X-ray pulses irradiating the object 11 is detected by means of the matrix 12 .

[0036] Radiation 13 may be generated by an X-ray source 15, which in some examples may be rigidly connected to the matrix 12. The source 15 may be movable with the matrix 12.

[0037] exist Figure 2 In the example shown, the inspection system 10 may be mobile and transportable from one location to another (the system 10 may include a motor vehicle). Alternatively or additionally, the inspection system may be stationary relative to the ground and may not be displaceable.

[0038] exist Figure 2 In the example shown, the system 10 may also include an electro-hydraulic boom 14 that is operable in a retracted position in a transport mode (not shown) and in an inspection position ( Figure 2 The boom 14 may be operated by a hydraulic actuator (eg, a hydraulic cylinder).

[0039] The matrix 12 may be mounted on a boom 14 facing a source 15 on the other side of the cargo 11 .

[0040] In some examples, scanning the cargo at S1 may include scanning the cargo 11 in a mobile mode. In this case, displacing the cargo 11 and the matrix 12 relative to each other may include keeping the cargo stationary relative to the ground during the scanning displacement SD and displacing the matrix 12 relative to the ground. In some examples, scanning the cargo may include mounting the matrix 12 on a cantilever 14 of the mobile inspection system 10. In this case, generating the corrected image at S5 may include correcting cantilever movement artifacts generated by movement of the cantilever 14 on which the matrix 12 is mounted in the mobile mode (e.g., parasitic displacements, such as vertical displacements, rotational displacements, and / or pendulum displacements).

[0041] Alternatively or additionally, scanning the cargo 11 at S1 may include scanning the cargo 11 in a pass-through mode. In this case, displacing the cargo 11 and the matrix 12 relative to each other may include keeping the matrix 12 stationary relative to the ground and displacing the cargo 11 relative to the ground during the scanning displacement SD. In some examples, scanning the cargo 11 includes mounting the matrix 12 on a cantilever 14 of the static inspection system 10. In this case, generating a corrected image at S5 may include correcting cargo speed non-uniformities (such as speed non-uniformities of the cargo 11) generated in the pass-through mode.

[0042] In this description, all figures depict the case of vertical boom movement, for example in the moving mode. It should be understood that for speed non-uniformity, for example in the pass-through mode, all explanations will remain similar with reference to similar figures with horizontal parasitic displacement instead of vertical parasitic displacement.

[0043] like Figure 4 As shown, in some examples, each region Zi of the cargo 11 irradiated by a first X-ray pulse Pi is irradiated by at least one second X-ray pulse Pj (j≠i).

[0044] As an alternative or in addition, Figure 4 As shown, radiation 13 corresponding to a first X-ray pulse Pi is detected by a first row 121 - i of the matrix 12 , and radiation 13 corresponding to at least one second pulse Pj (j≠i) is detected by at least one second row 121 - j (j≠i) of the matrix 12 .

[0045] In the examples of the present disclosure, the speed of the mutual scanning shift SD is such that between two pulses Pi and Pi+1, the mutual shift of the goods 11 relative to the matrix 12 can be less than the width of the row 121-i of the matrix 12 (e.g., about 2 mm, but other dimensions are conceivable). In some examples, any area Zi of the goods 11 illuminated by a pulse Pi will be detected by all rows 121-i of the matrix 12. Figure 4As shown, the matrix 12 may oversample the goods 11. In some examples, the area Z1 of the goods 11 appears several times, in different rows (e.g., Figure 4 121-1, 121-2, 121-3, 121-4 and 121-5 in the example).

[0046] like Figure 1 As shown, the method 100 may further include, at S2, generating a first image of the goods 11 for a first row 121-i of the matrix 12, and generating at least one second image of the goods 11 for at least one second row 121-j (j≠i) of the matrix 12. In some examples, an image may be generated for each row 121 of the matrix 12 from data provided by the matrix 12.

[0047] In some examples, each image includes pixels corresponding to rows 121 - i of detector 122 for each X-ray pulse Pi.

[0048] In some examples, such as Figure 5 As shown in FIG. 1 , a local mutual parasitic shift may occur between the first pulse Pi and the at least one second X-ray pulse Pj (j≠i). It should be understood that the local mutual parasitic shift is different from the mutual scanning shift SD (e.g., Figure 4 shown).

[0049] In some examples, the mutual scanning shift SD occurring between the first pulse Pi and at least one second pulse Pj (j≠i) may correspond to the case where the movement of the matrix 12 may be only a translation at a constant speed along the horizontal axis ( Figure 4 ). In such an example, the images generated may only be translated between them, this translation corresponding to the distance travelled by the matrix 12 relative to the goods 11 between the two pulses Pi and Pj (j≠i).

[0050] like Figure 5 As shown, the matrix 12 may have a mutual parasitic shift PD (for example, when the matrix 12 is mounted on a cantilever 14 of the inspection system 10) that is superimposed on the mutual scanning shift SD. Figure 5 In an example of , the mutual parasitic displacement PD may include a vertical displacement PDy. Other examples of mutual parasitic displacement may include rotational displacement and / or pendulum displacement and / or velocity non-uniformity.

[0051] For clarity, in Figure 5 In the example, the vertical parasitic movement PDy and the scanning shift SD between two pulses Pi are exactly one pixel.

[0052] In some examples, the speed of the vertical parasitic movement PDy can be considered to be Figure 5 The four pulses P1, P2, P3 and P4 shown in are constant in duration, since most parasitic movements have a frequency with a typical period of about one second (in the range of 0.5 to 1.5 seconds, depending on the inspection system 10). The frequency of the pulses Pi may be comprised, for example, between 100 Hz and 300 Hz, and may typically be 200 Hz. At a frequency of 200 Hz, there are about 200 pulses during one mechanical oscillation of the matrix 12.

[0053] Fig. 6A and 6B Examples of simplified images of different rows 121 - i are schematically shown. Fig. 6A An image is schematically shown without correction for the scan shift SD. The scan shift SD is known and can therefore be corrected. Figure 6B An image with correction for scan shift SD is schematically shown.

[0054] It should be understood that, since the speed is assumed to be Fig. 6A and 6B In the example of , the images for different rows 121-i of goods 11 are locally similar. This means that, locally, an image can be transformed into another image by a translation vector. The variation of the translation vector in time (i.e., along the horizontal direction) and / or in the vertical direction depends on the nature of the parasitic shift (e.g., vertical, rotational and / or oscillatory).

[0055] Therefore, the method 100 may further comprise, at S3, determining, for each region Zi of the cargo 11 irradiated by the first X-ray pulse Pi and by the at least one second X-ray pulse Pj (j≠i), a local mutual parasitic shift between the cargo 11 and the matrix 12 .

[0056] It should be understood that all different images are distorted by the parasitic shift PD and no image can be considered a reference. However, in some examples, determining the local mutual parasitic shift at S3 may include determining the local mutual shift between corresponding pixels of the first image and at least one second image. In some examples, instantaneous and local shift vectors may be determined and used to correct the parasitic shift in the image.

[0057] The following conventions will be used in the following developments:

[0058] I1(x,y) may be the image that would be acquired by the first row 121 - 1 if there were no parasitic shifts;

[0059] I n (x, y) may be the image that would be acquired by the nth row 121 - n if there were no parasitic shift;

[0060] x is the horizontal coordinate, i.e. the scanning direction (Ox); and

[0061] y is the vertical direction

[0062] I n (x,y) = I1 (x-(n-1).Δ,y) (Formula 1)

[0063] x and y are distances, and Δ is the distance between two adjacent rows 121 - i and 121 -i+l at the cargo 11 .

[0064] (n-1).Δ is a constant offset in the scan direction (Ox) and is known because the scan speed is known. In some examples, (n-1).Δ can be eliminated by shifting the image horizontally (and filling the missing rows with zeros at the end of the image). In some examples, I n (x,y) can be converted to I n (x+(n-1)Δ,y) is replaced.

[0065] The slight stereo effect can also be ignored, and the converted image can look like this:

[0066] I n (x,y)=I1(x,y).(Formula 1')

[0067] I n (x,y) can be an image obtained using parasitic shift. For the first row:

[0068] I1(x,y) = I1(x +Δ x (t1(x),y) , y+ Δ y (t1(x),y)) (Formula 2)

[0069] in:

[0070] t1(x) is the time of the first row of image coordinates x; and

[0071] Δ x (t1(x),y) and Δ y (t1(x), y) are the parasitic shifts in x and y, respectively, due to the relative movement of the matrices at time t1(x).

[0072] Formula 2 may be valid for continuous models (e.g., images may be functions of continuous variables). However, the cargo images are functions of x and y as discrete variables, and it can be assumed that the cargo images are bandwidth limited and the pixel size allows reconstruction of continuous functions, for example, using the Nyquist-Shannon theorem.

[0073] For the nth row, the constant translation due to the scan SD is suppressed:

[0074] I n (x,y)=I n (x+Δ x (t n (x),y),y+Δ y (t n (x),y))=I1(x+Δ x (t n (x),y),y+Δ y (t n (x),y))

[0075] in:

[0076] t n (x) is the time of the image coordinate x of the nth row; and

[0077] Δ x (t n (x), y) and Δ y (t n (x),y) is due to the matrix at time t n The image shift is caused by moving the matrix of (x).

[0078] For clarity, for I1, we can use t n (x) replaces t1(x) because the image is corrected from a constant scan shift SD. These instants are separated by nδ / v, v being the scan velocity.

[0079] Time and average position (i.e., position without the matrices shifting relative to each other) are proportional (x=vt, where v is the scan speed). Therefore, it is more convenient to remove the time and express Equation 2 as follows:

[0080] I n (x,y) = I1(x + Δ nx (x,y) , y+Δ ny (x,y)) (Formula 3)

[0081] The following development describes how to determine the local shift vector between any pair of images.

[0082] Figure 7 The shift field DF and the estimation of the instantaneous shift field between two images from two different detection rows (eg row 121 - 1 and row 121 - 5 ) are schematically shown.In the context of the present disclosure, the instantaneous shift is associated with the shift between two consecutive pulses Pi.

[0083] like Figure 7As shown, two images I corresponding to rows 121-p and 121-q of the matrix 12 can be estimated. p (x, y) and I q The shift field DF between (x, y) is such that:

[0084] Δ px (x, y)-Δ qx (x, y); and

[0085] Δ py (x, y)-Δ qy (x, y).

[0086] Assuming that the speed of the matrix shift between the two rows 121-i and 121-i+1 is constant, the instantaneous shift field IF can be determined by dividing the above shift field by pq using the first row 121-1 and the fifth row 121-5, as Figure 7 As shown, the simple case is a vertical shift only.

[0087] like Figure 1 As shown, the method 100 may further include, at S4, determining a total mutual parasitic shift by summing the determined local mutual shifts for the irradiated area Zi from an X-ray pulse associated with the start of the cargo scanning to an X-ray pulse associated with the end of the cargo scanning.

[0088] In some examples, the instantaneous shift field IF described above may be used to determine image distortion as described below.

[0089] like Fig. 8A and 8B As shown, according to the instantaneous shift δ on x and y x (x,y) and δ y (x,y), the actual positions x' and y' can be defined such that:

[0090]

[0091] The calculation in Equation 4 describes the mapping between the acquired image coordinates (x and y) and the actual position (x' and y'). The mapping can be performed horizontally, row by row, because the instantaneous shift δ is calculated between adjacent rows. x (x,y) and δ y (x,y). Fig. 8A and 8B The geometric representation of Equation 4 is schematically shown, Fig. 8A The coordinates of the acquired images are schematically shown. Figure 8B Schematically shows Fig. 8A The coordinates of the actual position of the acquired image. Fig. 8A and 8BIn , the matrix shift moves the gray level from (x′, y′) to (x, y).

[0092] exist Fig. 8A and 8B In , the horizontal straight line is transformed into a curve CL, which is also a function of x. Figure 8B As shown, the components of the tangent vector V of the curve CL are (1+δ x (x,y),δ y (x,y)). Note that x and y are asymmetric because the shift is not a real function of x and y, but a function of time t and y. However, time t can be replaced by x because time t and x are proportional. In (1+δ x (x,y)), 1 comes from the time stream.

[0093] In the above formula 4, the lower limit of the integral (sum) is zero, but it should be understood that the lower limit can be any value x0, that is, from the X-ray pulse associated with the start of the cargo scan. In this case, the equation is:

[0094]

[0095] In the above formula 4, the upper limit of the integral (sum) can be associated with a row that can be selected as a reference. It is necessary to determine the total displacement of each point of the image relative to its position without any displacement. Therefore, it is assumed that the reference row can correspond to the static position of the matrix. The reference row can be associated with the X-ray pulse associated with the end of the cargo scan.

[0096] Most of the theoretical development above was done using a continuous model. A more realistic discrete equivalent equation would be (taking the first column as a reference):

[0097]

[0098] X ij and ij are integers (since i and j are pixel coordinates), and x i ' j ,y i ' j is not an integer (in the general case, the parasitic shift in detector space is not a multiple of the line width).

[0099] In some examples, the gray levels of the integer pairing values ​​x' and y' are determined because they correspond to the coordinates of the pixel. There are several numerical ways to determine the gray levels of the integer pairing values ​​x' and y', for example using deformation. Fig. 9 This principle is schematically illustrated, where the reverse mapping is shown. Fig. 9In , for a given pixel (i', j'), the pixel (i, j) whose image overlaps (i', j') due to distortion can be determined. The gray level in (i', j') can be calculated by the sum of the gray levels of the pixels (i, j) weighted by the overlap ratio.

[0100] In the simplest case, the matrix shift can be approximated by a vertical translation (δ x =0,δ y Depends only on x):

[0101]

[0102] like Figure 1 As shown, the method 100 may further include generating a corrected image of the cargo 11 based on at least one of the first image and the at least one second image and based on the determined total mutual parasitic shift at S5.

[0103] In some examples, computing the corrected image may be performed by translating each column by an amount D equal to:

[0104]

[0105] As described above, the results of D above correct for cantilever motion which is primarily vertical translation (eg, the effects of rotation are believed to be very small).

[0106] In the general case, in x and y:

[0107]

[0108] Therefore, the P image can be corrected according to the following formula:

[0109] I cn (x,y)=I n (x-Dx(x,y),y-Dy(x,y))

[0110] Among them I cn (x,y) is the rectified nth image.

[0111] The final calculation may not be performed on all images, since the corrected images may be assumed to be similar.

[0112] As follows about Figure 2 As explained in further detail, the inspection system 10 is configured to inspect cargo 11 , which, in some non-limiting examples, may be performed by transmitting inspection radiation 13 from an inspection radiation source 15 through the cargo 11 to the matrix 12 .

[0113] exist Figure 2In the example shown, the system 10 may also include a controller 16, which is configured to communicate with the matrix 12 via a communication network that may be wired and / or wireless to generate one or more images based on data received from the matrix 12, and the controller 16 includes at least a processor and a memory so as to at least partially execute the example method 100 according to the present disclosure locally on the system 10, for example.

[0114] exist Figure 2 In the example shown, the system 10 may also include an analyzer 17, which is configured to receive one or more images from the controller 16, for example, via a communication network, which may be wired and / or wireless. The analyzer 17 includes at least a processor and a memory, so as to perform, for example, at least partially, an example method 100 according to the present disclosure remote from the system 10.

[0115] In some examples, determining the local mutual parasitic shifts at S3 may include using at least one of: an image cross-correlation technique and an energy minimization technique.

[0116] Cross-correlation of images is known to those skilled in the art for determining shift fields in a sequence of images. The cross-correlation technique will now be briefly described.

[0117] The shift vector can be found locally using the Fourier transform. p and I q are two images corresponding to rows p and q, then F p and F q is the 2D Fourier transform.

[0118] The normalized cross power spectrum R can be calculated such that:

[0119]

[0120] where ° is the matrix operation (entry-wise product) and * is the complex conjugation.

[0121] If I p and I q is the shift (δ x ,δ y ), then the cross power spectrum is:

[0122]

[0123] To get the cross-correlation, we can determine that now (δ x ,δ y ) is the inverse Fourier transform of the cross power spectrum of the Dirac delta function located in . Find (δ x ,δ y) may require calculating R and locating the maximum value of the Fourier transform of R.

[0124] To determine the shift field, the above method can be applied to each pixel (x, y) over a small square window (n × n pixels) centered at the pixel, and the result can be considered as a local shift (δ x (x,y),δ y (x,y)). The obtained local shift can be divided by (qp) to obtain the local instantaneous shift.

[0125] Energy minimization is also known to those skilled in the art.Energy minimization techniques will now be briefly described.

[0126] A convex or pseudo-convex function with the instantaneous displacement vector as a variable is established and the minimum of the function is determined. The usual basic term of such a function is the "consistency term". Always consider the I as described above p and I q (p>q), the first basic term is such that:

[0127] E C ((δ x ) ij ,(δ y ) ij )= ∑ ij (I p (i-(pq).(δ x ) ij ,j-(pq).(δ y ) ij )-I q (x,y) 2 (Formula 7)

[0128] For clarity, a discrete form of the function is used in Equation 7. The “consistency term” above can be the offset image (term I p ) and the target image (item I q ) is the square of the difference between . The algorithm mainly aims to determine the (δ x ) ij and (δ y ) ij The determined value corresponds to making the image I p Looks similar to image I q Shift values ​​as similar as possible.

[0129] Other terms of normalization can be added to Equation 7 to stabilize the obtained results:

[0130]

[0131] Here, the gradient is calculated as the delta between two adjacent pixels. x and δ y The coefficient a (positive) indicates that E T The higher a is, the smoother the transformation is. The shift to be determined is not an integer, and a warping / interpolation algorithm can be used to calculate the term Ec.

[0132] Under the constant velocity assumption as described above, the function for integrating all possible image pairs can be such that:

[0133]

[0134] Among them, E c (n,p,(δ x ) ij ,(δ y ) ij )=∑ ij (I p (i-(pn).(δ x ) ij ,j-(pn).(δ y ) ij )-I q (x,y) 2

[0135] Gradient descent techniques can be used to minimize E T .

[0136] exist Figure 2 In the example shown, the inspection radiation source 15 may include an X-ray generator. The energy of the X-rays may be between 100 keV and 15 MeV, and the dose rate may be between 2 mGy and 20 Gy (Gray) per minute at a distance of one meter from the source. Figure 2 In the example shown, for steel penetration capability of, for example, between 40 mm and 400 mm, typically 300 mm (12 inches), the maximum X-ray energy of the X-ray source 15 may be, for example, between 100 keV and 9.0 MeV, typically 2 MeV, 3.5 MeV, 4 MeV or 6 MeV. Figure 2 In the illustrated embodiment, the dose may be, for example, 20 mGy to 120 mGy. In other examples, for steel penetration, for example, between 300 mm and 450 mm, typically 410 mm (16.1 inches), the maximum X-ray energy of the X-ray source 15 may be, for example, between 4 MeV and 10 MeV, typically 9 MeV. In some instances, the dose may be 17 Gy.

[0137] The system 10 may also include other types of detectors, such as optional gamma and / or neutron detectors, for example, adapted to detect the presence of radioactive gamma and / or neutron emitting materials within the cargo 11, such as simultaneously with an X-ray inspection.

Claims

1. A method for inspecting cargo using X-rays, the method comprising: Scanning the cargo with a matrix comprising at least two rows of detectors, comprising: Utilizing mutual scanning displacement, the goods and the matrix are displaced relative to each other; and detecting, with the matrix, radiation produced by a plurality of successive X-ray pulses irradiating the cargo during the mutual scanning shifts; wherein each area of ​​the cargo irradiated by a first X-ray pulse is irradiated by at least one second X-ray pulse; and wherein radiation corresponding to the first X-ray pulse is detected by a first row of the matrix, and radiation corresponding to the at least one second pulse is detected by at least one second row of the matrix; generating a first image of the goods for the first row of the matrix and at least one second image of the goods for the at least one second row of the matrix, wherein each image comprises pixels corresponding to the row of detectors for each X-ray pulse; determining, for each region of the cargo irradiated by the first X-ray pulse and by the at least one second X-ray pulse, at least one local mutual parasitic shift between the cargo and the matrix by determining a local mutual shift between corresponding pixels of the first image and the at least one second image, the local mutual parasitic shift: occurs between the first X-ray pulse and the at least one second X-ray pulse; and different from the mutual scanning shift occurring between the first X-ray pulse and at least one second pulse; determining at least one total mutual parasitic shift by summing the determined local mutual shifts for the irradiated areas from an X-ray pulse associated with the start of the scanning of the cargo to an X-ray pulse associated with the end of the scanning of the cargo; and At least one corrected image of the cargo is generated based on at least one of the first image and the at least one second image and based on the determined at least one total mutual parasitic shift.

2. The method according to claim 1, wherein: Scanning the cargo includes: scanning the cargo in a moving mode, and shifting the cargo and the matrix relative to each other includes: During the scanning shift, the cargo is kept stationary relative to the ground, and the matrix is ​​shifted relative to the ground.

3. The method of claim 2, wherein scanning the cargo comprises: The matrix was mounted on a cantilever of a mobile inspection system.

4. The method according to claim 3, wherein: Generating the corrected image comprises: Cantilever movement artifacts generated in the movement pattern are corrected.

5. The method according to claim 1, wherein: Scanning the cargo includes: scanning the cargo in a pass-through mode, and shifting the cargo and the matrix relative to each other includes: During the scanning shift, the matrix is ​​kept stationary relative to the ground and the cargo is shifted relative to the ground.

6. The method according to claim 1, wherein: Scanning of the goods includes: The matrix was mounted on a cantilever of a static inspection system.

7. The method according to claim 5, wherein: Generating the corrected image comprises: Corrects for uneven cargo speeds in pass-through mode.

8. The method of claim 1 , wherein determining the local mutual parasitic shift comprises: At least one of an image cross-correlation technique and an energy minimization technique is used.

9. The method according to any one of claims 1 to 8, wherein: Generating the at least one corrected image at a pixel (x, y) comprises: Translate each pixel (x, y) by the amount corresponding to the total mutual parasitic shift - D x (x,y) and -D y (x,y), such that: Among them, Δ x (x,y) and Δ y (x, y) are the local parasitic shifts in the horizontal scanning direction (Ox) and the vertical direction (Oy).

10. The method according to claim 9, wherein: include: Determine two images I corresponding to rows p and q of the matrix p (x, y) and I q The shift field between (x, y) is such that: Δ x ((p-q),y)=[Δ px (x,y)-Δ qx (x,y)] / (p-q) Δ y ((p-q),y)=[Δ py (x,y)-Δ qy (x,y)] / (p-q), Among them, Δ x ((pq),y) and Δ y ((pq), y) are the local parasitic shifts of the two images corresponding to row p and row q in the horizontal scanning direction (Ox) and the vertical direction (Oy), Δ px (x, y)-Δ qx (x, y) and Δ py (x, y)-Δ qy (x, y) are the shift fields between the two images corresponding to row p and row q respectively.

11. The method of claim 9, wherein in the case where the parasitic movement is a vertical translation movement, -D x (x,y)=0 and -D y (x,y)=-D so that: in, δ y =Δ y (x,y),Δ y (x,y) depends only on x, i is the pixel coordinate on the x-axis.

12. The method according to claim 9, wherein: Generating the at least one corrected image comprises: Generate the corrected nth image I cn (x,y), such that: I cn (x,y)=I n (x-D x (x,y),y-D y (x,y)), Among them, I n (x, y) is the nth image generated without parasitic shift.

13. The method according to claim 12, wherein: Generating the corrected nth image includes: Only a subset of corrected images equal to or less than the number of rows of the matrix is ​​generated.

14. An apparatus comprising: at least one processor; as well as Memory, The memory comprises instructions which, when executed by the processor, enable the processor to perform the method according to any one of claims 1 to 13.

15. A computer program product comprising instructions which, when executed by a processor, enable the processor to perform the method according to any one of claims 1 to 13 or to provide an apparatus according to claim 14.

Citation Information

Patent Citations

  • Radiation image correction method, device and system

    CN107680065A

  • Scanner linearity tester

    CN1621946A

  • X-ray baggage inspection apparatus

    JP1998185841A

  • Radiographic system and method for generating radiation phase contrast image

    JP2012120650A

  • Live Fluoroscopic Roadmapping Including Targeted Automatic Pixel Shift for Misregistration Correction

    US20080027316A1