Dual-energy detector and detector data processing method

Abstract

A dual-energy X-ray detector (100) is disclosed, having a first detector row (110) with first detector elements (111) and a second detector row (120) with second detector elements (122) arranged in parallel thereto, wherein the detector rows (110, 120) are arranged in parallel to each other in a row direction and in front of each other in the direction of X-ray beams (RX) to be detected, such that the projections of the first and second detector rows (110, 120) in the direction of one of the X-ray beams (RX) to be detected, which passes through the surface barycenter of a reference detector element of the first or second detector row (110, 120), are offset in overlap with each other by an effective offset (Δx; Δy). An X-ray examination apparatus (200) comprising such a detector (100) and a method for processing detector data provided by means of the detector (100) are also disclosed.

Description

Technical Field

[0001] This invention generally relates to the spatial resolution of X-ray images when imaging and non-destructively inspecting objects using dual-energy X-ray radiography to detect target objects, and more specifically to measures for improving the spatial resolution of the resulting X-ray images. In particular, this disclosure relates to dual-energy X-ray detectors with improved spatial resolution, X-ray inspection apparatuses including such detectors, and methods for processing detector data acquired by the detectors. Background Technology

[0002] The following introductory background information in this disclosure is intended only to provide a better understanding of the relationships described below, and represents prior art only to the extent that the cited literature contains.

[0003] It is well known to use dual-energy X-ray radiography to distinguish materials, and the basic considerations are described, for example, in the following literature: S. Kolkoori et al, "Dual High-Energy X-ray Digital Radiography for Material Discrimination in Cargo Containers," 11th European Conference on NonDestructive Testing, 2014, Prague, Proceedings; and Polad M. Shikhaliev, "Material Decomposed Cargo Imaging with Dual Energy Megavoltage Radiography," arXiv:1709.10406 ​​[physics.ins-det].

[0004] Jer Wang CHAN et al, "Wire transfer function analysis for castellated dual-energy x-ray detectors," APPLIED OPTICS, Volume 43, Number 35, December 10, 2004, Pages 6413-6420. Figure 1(a) shows a dual-energy x-ray detector array with one high-energy detector element and one low-energy detector element per pixel, arranged one on top of the other in the direction of the X-ray to be detected. EP1010021B1 shows a dual-energy X-ray detector line in which the line with the high-energy detector element and the line with the low-energy detector element are arranged one after the other in the scanning direction, so that the two lines can detect incident X-rays independently of each other.

[0005] In an X-ray inspection apparatus, the object to be inspected is conveyed through a scanning device at a predetermined conveying speed. The scanning device consists of detector rows orthogonal to the conveying direction and X-ray fans facing the detector rows. The object to be inspected is scanned line by line through the X-ray fans. The resolution of the resulting X-ray image is determined essentially by the area of ​​each detector element and their number per unit length or per unit area of ​​the detector row, as well as the ratio of the read-out frequency of the detector elements to the conveying speed of the object in the conveying direction, wherein each detector element corresponds to one pixel.

[0006] Because dual-energy X-ray radiography requires one low-energy detector element and one high-energy detector element per pixel, the entire detector requires twice the number of detector elements to achieve the desired spatial resolution. Therefore, dual-energy detectors already have a correspondingly high system cost. To achieve higher spatial resolution in the resulting X-ray image, the number of detector elements per unit length or region can be increased accordingly. However, this leads to a corresponding increase in the system cost of the detector, and because increasing the density of detector elements requires a corresponding reduction in the area of ​​each detector element, it results in a deterioration in the signal-to-noise ratio in the acquired detector data.

[0007] As an alternative to increasing the number of detector elements, the spatial resolution of the X-ray image in the scanning direction corresponding to the transmission direction can be achieved in the aforementioned X-ray inspection apparatus by reducing the transmission speed or increasing the readout frequency. The former has an adverse effect on the throughput of the object being inspected at the X-ray inspection apparatus, while the latter further deteriorates the signal-to-noise ratio.

[0008] US8433036B2 discloses a method, system, and detector apparatus for scanning an object moving in a first direction, wherein a first detector region having a thickness of at least 2 mm and a second detector region having a thickness of at least 5 mm are provided, the second detector region being arranged to receive radiation that has passed through the first detector region.

[0009] US9329301B2 discloses a radiation detection device for foreign object inspection using a subtraction method, wherein a first radiation detector detects radiation in a first energy range passing through a sample, a second radiation detector detects radiation in a second energy range that is higher than the radiation in the first energy range, and wherein the thickness of a first scintillator layer of the first radiation detector is less than the thickness of a second scintillator layer of the second radiation detector, and wherein a first region of each pixel in a first pixel portion of the first radiation detector is less than a second region of each pixel in a second pixel portion of the second radiation detector.

[0010] US10386502B2 discloses a dual-energy detector and a radiation testing system. The dual-energy detector includes: a detector module holder; and a plurality of detector modules. Each detector module includes a higher-energy detector array and a lower-energy detector array arranged side-by-side on the detector module holder to be irradiated independently of each other.

[0011] US 2019 / 0179038A1 discloses an X-ray image sensor having two or more energy regions as part of an X-ray object detector, said energy regions being implemented as a side-by-side pixel array on a planar monolithic semiconductor substrate. Each pixel array in this monolithic side-by-side arrangement is designed to respond to a specific X-ray energy range or spectrum. Summary of the Invention

[0012] The objective of this invention is to provide a dual-energy X-ray detector or an X-ray inspection apparatus including such a dual-energy X-ray detector, which enables higher spatial resolution in the generated X-ray images.

[0013] Preferably, for this purpose, the number of detector elements per unit length or area should not be increased and / or the scanning speed of the X-ray inspection apparatus on the object to be inspected should not be reduced.

[0014] For example, if, with a constant number of detector elements (at least high-energy X-ray detection, hereinafter referred to as "high-energy") Hi(gh) and (at least low-energy X-ray detection, hereinafter referred to as "low-energy") Lo(w), it is possible to derive an X-ray image with higher spatial resolution in an image direction orthogonal to and / or along the scanning direction from the detector data detected by the detectors, it would be an improvement to a dual-energy detector.

[0015] The aforementioned task can be accomplished using the features of claim 1 relating to a dual-energy X-ray detector. Further embodiments and further embodiments are defined in the following dependent claims.

[0016] The inventors have recognized that a defined effective offset between the still overlapping effective regions of the Hi(gh) detector elements in a Hi detector line and the Lo(w) detector elements in a Lo detector line can be advantageously used to improve spatial resolution, in a conventional dual-energy X-ray detector where one of a Hi detector element and one of a Lo detector element are precisely assigned to the same pixel. This effective offset can be configured in the direction of the detector line and / or orthogonal to the detector line; the corresponding possibilities achievable using this method are further explained below. Therefore, without increasing the total number of detector elements, the spatial resolution of the detector can be correspondingly increased in the direction of the detector line and / or in a direction orthogonal to the detector line.

[0017] It should be noted that, in the context of this disclosure, "effective offset" is intended to mean that the offset is not only an actual local offset between the effective regions of the Hi detector elements of the Hi detector row and the Lo detector elements of the Lo detector row, but also, alternatively or additionally, can be achieved by specifically aligning the detectors composed of the Hi detector rows and the Lo detector rows relative to the X-rays guided onto the detectors during operation, without necessarily being locally offset from each other. Of course, these measures can also be used in combination.

[0018] The core idea of ​​the line-shaped dual-energy X-ray detector proposed here is to design the detector structure and / or arrange the detector rows relative to each other and relative to the X-ray source such that, in operation, the projections of the Hi detector elements of the Hi detector row and the Lo detector elements of the Lo detector row are offset relative to each other in an overlapping manner in the direction of the reference X-ray beam, relative to the beam path of the X-rays emitted from the X-ray source to the detector; this "reference X-ray beam" can be, for example, an X-ray beam passing through the area centroid of the detector row or the area centroid of the reference detector element of the detector row. The reference detector element can be the detector element at the center of the detector row, thus having approximately equal distances from all edges of the detector row. As a result, compared to detectors with conventional arrangements and alignment with the X-ray generator or reference X-ray beam, using the new detector in the acquired dual-energy detector data (to be evaluated in order to derive associated X-ray images) no longer allows the Hi detector data and / or Lo detector data pairs to be allocated in a 1:1 ratio, i.e., uniquely assigned to exactly one image point. That is, at the location of the Lo detector element, there is a clear lack of a 1:1 allocatable Hi detector element, and vice versa.

[0019] The “region centroid of a detector row” here refers to the geometric centroid of a geometric figure formed by the effective regions of a type of detector element, whose region centroid corresponds to the average of all points within the figure. In the case of detector rows that are typically rectangular in shape, the region centroid of the detector row, or the region centroid of the reference detector element of the detector row, can be determined as the intersection of the lines that bisect the effective regions of the detector row in the longitudinal and width directions, or is usually calculated mathematically by integration.

[0020] The inventors have recognized that, due to the effective offset of Hi and / or Lo detector elements, missing detector data of one type for each real Hi and / or Lo detector data datum, which can no longer be assigned 1:1 to each other, can be supplemented by another type of virtual detector data calculated for that location. That is, based on a predetermined effective overlap offset between the Hi detector elements of the Hi detector row and the Lo detector elements of the Lo detector row, virtual detector data can be calculated with sufficient accuracy for replacement of the missing detector element at its location. This calculation has been found to be surprisingly feasible, and the X-ray images derived from the real and virtual detector data show practical improvements.

[0021] According to a first aspect, a linear dual-energy X-ray detector (hereinafter referred to as the detector) is provided for an X-ray inspection apparatus (e.g., an X-ray inspection apparatus according to the second aspect described below), the X-ray inspection apparatus being configured to perform a dual-energy X-ray radiography method on an object to be inspected.

[0022] In the context of the detector proposed in this paper, "linear" should initially be understood as meaning that the detector is configured to acquire detector data for multiple image points in the longitudinal direction of the detector and for a smaller number, but at least one, image point orthogonal to the longitudinal direction. In the simplest implementation, the detector thus has two detector rows, each with an equal number of Hi(gh) and Lo(w) detector elements in the longitudinal direction, and each detector row has a single Hi and a single Lo detector element orthogonal to it. In principle, the detector can also have multiple Hi and Lo detector elements in the direction orthogonal to the longitudinal direction; therefore, the principle proposed here is applicable.

[0023] The proposed linear detector comprises at least: a first detector row having a first detector element; and a second detector row arranged parallel to the first detector row and having a second detector element. The detector rows are configured to respond selectively, differently in spectral terms, to the spectrum of the X-ray to be detected. The detector rows are arranged parallel to each other in a row direction. The detector rows are arranged one after another in the direction of the X-ray to be detected, such that the projections of the first and second detector rows in the direction of one of the X-rays to be detected (which serves as a reference X-ray beam) are offset and overlap. The projections of the first and second detector rows may be in the direction of one of the X-rays to be detected (which passes through the centroid of the region of the reference detector element of the first or second detector row). The reference X-ray may be an X-ray that passes through the centroid of the entire effective region of the detector element of the foremost detector row relative to the X-ray to be detected, or an X-ray that passes through the centroid of the effective region of the reference detector element of the foremost detector row relative to the X-ray to be detected.

[0024] The first detector row may have a relevant first or Lo(w) detector element for primarily detecting low-energy X-rays, and the second detector row may have a relevant second or Hi(gh) detector element for primarily detecting high-energy X-rays.

[0025] The Lo and Hi detector elements can be configured to generate their respective associated Lo detector data or Hi detector data. In principle, the Lo and Hi detector elements can be substantially identical, as the Lo detector element typically absorbs more low-energy quanta than the Hi detector element below it because it is generally closer to the X-ray source. Therefore, the superposition of the Lo and Hi detector elements already results in a (small) split in the X-ray spectrum. To amplify this split, a filter material for reinforcement (i.e., filtering of low-energy X-rays) can be additionally disposed between the first and second detector elements.

[0026] In a first embodiment, the first detector row and the second detector row are offset from each other by a first (local) offset in the row direction. Using this first embodiment, an increase in the spatial resolution of the X-ray image derived in the row direction (i.e., laterally to the scanning direction) can be achieved.

[0027] In the second embodiment, the first detector row and the second detector row are offset from each other by a second (local) offset orthogonal to the row direction. Using the second embodiment, the spatial resolution of the X-ray image to be derived orthogonal to the row direction (i.e., in the scanning direction) can be increased. As a result, using the second embodiment, the spatial resolution of the derived X-ray image can be increased in the scanning direction without reducing the transport speed of the object being examined or the ratio of transport speed to scanning frequency. In an alternative application of the second embodiment, the transport speed of the object being examined or the ratio of transport speed to scanning frequency can be increased, and the throughput of the X-ray examination apparatus can be increased accordingly, while the spatial resolution of the derived X-ray image remains the same. A combination of these two effects is also possible with appropriate configurations.

[0028] The first and second embodiments can be used alternatively or in combination, i.e., used simultaneously.

[0029] As an alternative or supplement to the first and second embodiments described above, wherein the first and second detector rows are (partially) offset from each other, in the third embodiment, the first and second detector rows may also be tilted at a corresponding tilt angle relative to the reference X-ray beam to achieve effective offset.

[0030] In this way, the desired effective offset can also be achieved using conventional detectors, in which the first and second detector elements can be clearly allocated in a 1:1 ratio, i.e., without local offset from each other. Of course, a third embodiment can be used in combination with the first and / or second embodiments to achieve the effects of the corresponding other embodiments; that is, the third embodiment can be used to achieve the effects of both the first and second embodiments.

[0031] Regarding the third embodiment, it should be noted that in conventional detectors, the first and second detector rows are typically arranged in series, such that the connecting line between the Lo detector element and the associated Hi detector element points precisely to the focal point of the associated X-ray generator aligned with the detector. Since the Lo and Hi detector elements arranged sequentially are spaced a predetermined distance from each other (e.g., approximately 1 cm) as viewed from the X-ray beam generator, the entire detector can be easily tilted at an appropriate angle (e.g., 5°–6°) relative to the reference X-ray beam in the longitudinal direction and / or in a direction orthogonal to it. This tilting of the detector causes the connecting line between the centroids of the regions of a pair of Hi detector elements and Lo detector elements to pass through the X-ray generator.

[0032] For example, when the detector is tilted longitudinally, the tilt can be specifically adjusted so that the line connecting the centroid of the region of the Lo detector element and the gap between two adjacent Hi detector elements points towards the X-ray generator.

[0033] For example, when the tilt is orthogonal to the longitudinal direction of the detector, the tilt can be specifically adjusted so that the connecting line between the centroid of the Lo detector element and the edge of the Hi detector element located behind it points to the X-ray generator.

[0034] Therefore, by utilizing the third embodiment, it is possible to produce an effect substantially the same as the linear (local) offset of the first and second detector rows in the longitudinal direction of the detector and / or in a direction orthogonal to the longitudinal direction (hence the "effective offset").

[0035] As a result, by utilizing the first, second, and third embodiments described above, a preferred and effective offset between the first and second detector elements can be achieved to achieve any of the above effects or combinations thereof.

[0036] In particular, by utilizing all three implementation methods or combinations thereof, the effective offset can be adjusted such that, in operation, for a detector pair formed by a real detector element and a calculated virtual detector element, the X-ray generator is aligned with the line connecting the centroids of the relevant regions of the two detector elements, i.e., focused on the focal spot of the generator.

[0037] In this respect, it should be noted in the following description that, in practice, the I-shaped, L-shaped, or U-shaped detector rows in an X-ray inspection apparatus can be formed by multiple individual linear detectors, each individually aligned relative to a reference X-ray beam on a corresponding associated X-ray generator of the apparatus. Therefore, in the third embodiment, the projections of the first and second detector rows along the reference X-ray beam are actually shifted relative to each other at least at the ends of the detector rows (approximately 14-15%) due to the geometry. Therefore, it is recommended that the overlap between the detector elements of the first and second detector rows be at most 80% and at least 20%.

[0038] The first and second detector elements can have substantially the same cross-sectional shape and cross-sectional area (effective area) in the direction of the X-rays to be detected. In other words, for a conventional arrangement where they are placed one after the other, the first and second detector elements can be congruent.

[0039] The “effective area” of a detector element is understood here as the effective area of ​​the detector element in which the X-rays to be detected, which have passed through the object being inspected, are incident on the effective area in the intended arrangement.

[0040] The first detector element and the second detector element each have a length along the row direction and a width orthogonal to the row direction.

[0041] In a particularly preferred first embodiment, the first offset in the row direction corresponds to half the length of the first and second detector elements. As a result, the preferred first embodiment provides a detector with twice the spatial resolution in the row direction.

[0042] In a particularly preferred second embodiment, the second offset is determined according to formula (1) based on the readout frequency f in [1 / s] for reading detector data from the detector element and the transmission velocity b of the object being examined relative to the dual-energy X-ray detector in cm / s:

[0043] Δz = (mb) / (2f) (1),

[0044] Where m is an odd integer (m = 1, 3, 5, 7, ...).

[0045] As a result, the preferred second embodiment provides a detector with twice the spatial resolution perpendicular to the row direction, and the ratio of the speed at which the object passes through the detector to the readout frequency of the detector element remains constant. Alternatively, using the detector according to the preferred second embodiment, the scanning speed for the object in an X-ray inspection apparatus with the detector can be correspondingly reduced, possibly halved, while maintaining the same spatial resolution. That is, with the setup according to the second preferred embodiment, optionally, a higher belt speed with constant image quality or improved spatial resolution with a constant belt speed is possible. In principle, the ratio of belt speed to scanning frequency can be adjusted as needed. In a particularly preferred second embodiment, if the belt speed and sampling frequency remain the same, the spatial resolution is improved using a new detector. If the belt speed remains constant and the sampling frequency is reduced, the resolution remains unchanged, but the signal-to-noise ratio is improved, making it possible to better resolve fine structures such as fine lines in X-ray images, for example. If the scanning frequency remains constant and the belt speed increases, the resolution also remains unchanged, but the throughput of the object per unit time in the system increases. Therefore, the preferred second embodiment provides particularly high flexibility, allowing the system to be designed for different needs.

[0046] Without detector offset, the distance traveled by baggage items on the conveyor belt between two readout processes is Δs = b / f, and from the measured values ​​Lo1, Hi1, Lo2, Hi2, Lo n Hi n ,...,Lo N Hi N Generate ((Lo1,Hi1),(Lo2,Hi2),...,(Lo N Hi N Value pairs in the form of )).

[0047] With an offset Δz = (mb) / (2f) and m = 1, as described above, the spatial resolution can be doubled in the transport direction because an intermediate value is generated at a distance Δs / 2. In this case, the value pairs are in the form of ((Lo1,vHi1),(vLo1,Hi1),(Lo2,vHi2),(vLo2,Hi2)...,(Lo N ,vHi N ),(vLo N Hi N At half the band speed (or double the readout frequency), Δs is halved, and this results in the detector data still being in the order Lo1, Hi1, Lo2, Hi2, Lo at the same offset Δz. n Hi n ,...,Lo N HiN The measurements were generated in the order of (Lo2,Hi1), but now Lo1 and Hi2 are spatially consistent. The measurements can now be assigned accordingly, for example, in ((Lo2,Hi1), (Lo3,Hi2), ..., (Lo...). N Hi N-1 The combination of these elements allows for operation without calculating intermediate values. That is, in this preferred second embodiment, the spatial offset between the associated first and second detector data caused by serial readout can be compensated for by a clever combination of the spatial offset of the detector element in a direction orthogonal to the row direction and the ratio of readout frequency to tape speed.

[0048] As described above, (depending on the target orientation) the two particularly preferred embodiments can be used independently or in combination. As described above, based on the effects and possible applications explained above, the third embodiment can be used alternatively or additionally to the first and / or second embodiments.

[0049] Therefore, in a particularly preferred first or corresponding third embodiment, the high-energy Hi detector element and the low-energy Lo detector element are effectively moved by half the number of detector elements compared to a conventional arrangement. In an alternative or additional preferred second or corresponding third embodiment, the ratio of readout frequency to tape speed, and correspondingly the effective ratio in the transmission or scanning direction, can be adjusted as needed as described above.

[0050] Although the Hi and Lo detector elements are configured to respond differently (spectrally selectively) to the incident X-rays to be detected, the readout frequency and band speed, as well as the known ratio of readout frequency and band speed, can be adjusted as needed due to the defined effective offset of the Hi and Lo detector elements in the row direction and / or orthogonal to the row direction. If necessary, the Hi detector data missing from the real Lo detector elements can be calculated as virtual Hi detector data based on the acquired adjacent real Hi detector and / or real Lo detector data. The possibility of the specific process is further explained below in conjunction with the method according to the third aspect for processing the first and second detector data obtained using the detector of the first aspect.

[0051] Regarding the implementation of the first and second detector elements, it should be noted that, in principle, the first and second detector elements can be constructed in a manner known per se for the desired spectral selectivity of the X-ray spectrum used. For example, each of the first and second detector elements may include a photodiode with a scintillator material disposed in front of it relative to the incident X-ray to be detected. The scintillator material can be selected to adjust the desired spectral selectivity sensitivity to the X-ray spectrum used, and / or the thickness of the scintillator material through which the X-ray passes can be appropriately determined. For example, YAG (yttrium aluminum garnet), YGAG (yttrium gadolinium aluminum garnet), GOS (gadolinium oxysulfide), or materials physically similar to X-rays can be used as scintillator materials. For low-energy scintillators, a layer thickness on the order of 1 mm is preferred, while for high-energy scintillators, another layer thickness on the order of 1 mm is preferred. The photons generated by the incident X-rays in the respective scintillator material are detected by the associated photodiode and converted into an output voltage, which depends on a time interval after which, in each case, the output signal generated by the photodiode is read out. The output signal read out in this regular manner corresponds to the relevant detector data.

[0052] Between the first and second detector elements, which are Lo-detector elements, a filter for enhancing the incident X-rays can be provided, such as a filter made of copper, titanium, aluminum, etc., which attenuates low-energy X-rays particularly strongly, so that the second detector element is mainly reached by high-energy X-rays, and is therefore a Hi-detector element.

[0053] According to one or a combination of the first and second embodiments, the first and second detector elements may be arranged on the same side of the carrier to form the detector proposed herein. Alternatively, the first detector element may be located on one side of the carrier, and the second detector element may be located on the other side of the carrier. For example, a printed circuit board (PCB) may be used as the carrier, on which conductor traces and, if necessary, functional electronics for contacting and reading out the detector elements are located.

[0054] Compared to conventional designs of known dual-energy detectors, a predetermined effective overlap offset between Hi- and Lo- detector elements arranged one after another in the direction of the X-ray to be detected is achieved at the same system cost.

[0055] The dual-energy detector proposed here achieves higher spatial resolution without increasing the number of Hi and Lo detector elements per unit length or unit area.

[0056] Compared to setups that use detector elements with smaller effective areas to increase the number of Hi and Lo detector elements per unit length or unit area, the dual-energy detector proposed in this paper has a better signal-to-noise ratio.

[0057] The dual-energy detector presented here is particularly simple. In principle, no significant modifications to existing detectors are required to put this concept into practice. In principle, the Hi and Lo detector elements need to be moved by a desired, predetermined relative offset of effective overlap, for example, preferably by half the length and / or width of the detector elements.

[0058] A second aspect of this disclosure provides an X-ray inspection apparatus comprising a dual-energy line-scan X-ray detector (detector) according to the first aspect.

[0059] The X-ray inspection apparatus is configured to transport the object to be inspected through the apparatus in a transport direction. The row direction of the dual-energy X-ray detector is preferably arranged orthogonally to the transport direction, such that the transport direction corresponds to the scanning direction for the object to be inspected. The X-ray inspection apparatus provides first detector data acquired in the form of Lo detector data and second detector data in the form of Hi detector data for the object to be inspected.

[0060] The X-ray inspection apparatus of the second aspect can perform dual-energy X-ray radiographic imaging for non-destructive inspection of an object and provides Lo and / or Hi detector data describing the object. The Lo and Hi detector data is based on the detection of X-rays on the object using the dual-energy X-ray detector of the first aspect. For this purpose, the object is passed through a scanning device at a predetermined or optionally set transport speed, the scanning device consisting of a linear detector arranged orthogonally to the transport line and X-ray fans aligned with the detector rows, for corresponding line-by-line scanning through the X-ray fans at a readout frequency of the detector elements, also predetermined or optionally set. As mentioned above, the ratio of transport speed to readout frequency can be adjusted as needed (high throughput or higher spatial resolution).

[0061] For the following description, the detector data acquired and provided by the detector is indexed by a position variable n, which extends from one end in the row direction of the detector. The first detector element, namely the Lo detector element, acquires the first real detector data of the inspected object, i.e., the real Lo detector data Lo1, Lo2, Lo... n ,...,Lo N Where 1 ≤ n ≤ N, and N is the number of Lo detector elements along the row direction of the detector. The second Hi detector element, i.e., the Hi detector element, acquires the corresponding second real Hi detector data of the inspected object, i.e., the real Hi detector data Hi1, Hi2, Hi3, ..., HiN The formula also maintains 1≤N≤N, where N is the number of detector elements in the detector row direction.

[0062] The inventors have also discovered that, in X-ray inspection apparatuses with conventional detectors, a similar effect on the ratio of achieved spatial resolution or transmission speed to the readout frequency of detector elements can be achieved by using specific readout modes for Hi- and / or Lo- detector data at conventional dual-energy X-ray detectors, as achieved in the second or corresponding third embodiment of the detector according to the first aspect. Specifically, the readout method described below uses a row of dual-energy X-ray detectors, each pixel having a high-energy Hi detector element and a low-energy Lo detector element, which are arranged substantially in the same orientation, one above the other, in the direction of the X-rays to be detected.

[0063] To describe the readout process in detail, as described above, the actual readout Hi and / or Lo detector data of the first and second detector rows are indexed: Hi or Lo represents the corresponding Hi detector data, which has a position variable n. For the corresponding position of the relevant detector element in the detector row, the position variable n = 1, 2, 3, ..., N, where 1 ≤ n ≤ N, and N is the number of detector elements of the relevant type (Hi or Lo) in the row direction.

[0064] The inventors have discovered that, using a readout sequence, all detector elements of one type (Hi or Lo) are initially read out along the position variable n, and then all other detector elements of other types are read out along the position variable n; that is, initially all Hi detector data Hi1, Hi2, Hi3, ... Hi N Then all Lo detector data Lo1, Lo2, Lo3, ..., Lo N (Or vice versa), the same result can be achieved as that achieved according to the second embodiment by shifting the first and second detector rows relative to each other. Therefore, the aforementioned readout method is a software solution equivalent to the hardware solution of the third embodiment according to the first aspect, in which Hi detector elements and Lo detector elements with effective offsets overlap each other.

[0065] Therefore, another aspect of the present invention relates to a method for reading out Hi- and Lo- detector elements for a dual-energy X-ray detector, and also to an X-ray inspection apparatus having the features of the X-ray inspection apparatus of the second aspect but having the following conventional detector, comprising at least one dual-energy X-ray detector row, each pixel having a high-energy Hi-detector element and a low-energy Lo-detector element arranged substantially uniformly (1:1) vertically in the direction of the X-ray to be detected, wherein the Hi- and / or Lo-detector data of the detector row to be read out is defined as Hi n or Lo n The method has a position variable n, where for the corresponding position of the relevant detector element in the detector row, the position variable n = 1, 2, 3, ..., N, where 1 ≤ n ≤ N, and N is the corresponding number of Hi- and Lo- detector elements in the row direction. The readout method includes: reading out the Hi- and Lo- detector elements such that all detector elements of one type are initially read out along the position variable n, and subsequently all other detector elements of another type are read out along the position variable n.

[0066] The third aspect of this disclosure relates to a method for processing Lo detector data and Hi detector data provided by the X-ray inspection apparatus of the second aspect or the conventional detector X-ray inspection apparatus described above and the readout method described above, such that an X-ray image with improved spatial resolution can be obtained from the processed detector data.

[0067] As described herein, the Hi- and / or Lo- detector data of each pixel acquired using the detector of the first aspect or the X-ray inspection apparatus of the second aspect, or an X-ray inspection apparatus having a conventional detector and the above-described readout method, no longer have a direct 1:1 relationship. However, the inventors have discovered that, due to a predetermined overlap offset between the Hi detector element and the Lo detector element, for example, Hi detector data missing from the real Lo detector element can be calculated as virtual Hi detector data, and correspondingly, Lo detector data missing from the real Lo detector element can be calculated as virtual Lo detector data.

[0068] The third aspect of the method essentially involves calculating the corresponding virtual Lo detector data at the location of the actual Lo detector element. It should be understood that, alternatively or additionally, the corresponding virtual Hi detector data can also be calculated at the location of the actual Lo detector element.

[0069] For example, calculating virtual Lo detector data at the location of a real Lo detector element may include: calculating virtual Lo detector data based on a first number of neighboring real Lo detector data and a second number of neighboring real Lo detector data. Therefore, alternative calculations of virtual Lo detector data may include: at the location of a real Lo detector element, calculating virtual Lo detector data based on a first number of neighboring real Lo detector data and a second number of neighboring real Lo detector data.

[0070] The inventors have discovered that various methods can be used to compute virtual Hi- and / or Lo- detector data, and these methods can be used alone or in combination.

[0071] For example, virtual Lo and / or Hi detector data can be calculated based on the average or median of adjacent Hi and / or Lo detector data. For example, Hi detector data lost from a Lo detector element can be calculated as virtual Hi detector data in a first manner as follows: based on a certain first number (e.g., two) of adjacent real Hi detector data (i.e., detector data of two Hi detector elements that overlap with the Lo detector element, i.e., those adjacent to it in the left-right direction in the row) and a certain second number (e.g., three) of the closest real Lo detector data (i.e., detector data of the considered Lo detector element and the two Lo detector elements adjacent to it in the upper left and upper right directions).

[0072] For example, the behavior of Hi detector data values ​​can be considered to calculate virtual Lo detector data, and / or the behavior of Lo detector data values ​​can be considered to calculate virtual Hi detector data.

[0073] The above methods can be implemented using machine learning algorithms. For example, deep learning algorithms can be used to generate virtual Hi and / or Lo detector data based on the provided Lo and / or Hi detector data.

[0074] Surprisingly, the inventors have discovered that methods from the field of processing image data from digital image sensors using so-called Bayer color filters are suitable for computing virtual detector data. In the case of optical image sensors with Bayer color filters, the color information lost at pixel locations can be computed using a so-called demosaicking algorithm (i.e., mosaic removal algorithm). For example, such an algorithm is described in the following literature: "Color filterarray demosaicking using high-order interpolation techniques with a weightedmedian filter for sharp color edge preservation" by JSJ Li and S. Randahawa, IEEE Transactions on image processing, Vol. 18, No. 9, September 2009; for reference, this literature is hereby abbreviated as Li / Randahawa, the contents of which are incorporated herein by reference.

[0075] In an image sensor with a Bayer color filter, the color filter elements of the three primary colors of red, green, and blue are arranged in what is called a Bayer matrix (or Bayer pattern) (see Figure 1 by Li / Randahawa), where the individual color filters are arranged in a checkerboard pattern, with 50% of the color filter elements being green and 25% of each color filter element being red and blue. The “Bayer matrix” and “Bayer filter” are named after its inventor, Bryce E. Bayer, and are described, for example, in US3971065A.

[0076] To apply this method, detector data acquired and provided by the dual-energy X-ray detector of the first aspect is processed via the aforementioned position variable n (which extends from one end in the row direction of the dual-energy X-ray detector) and indexed such that the Lo detector elements represent the actual Lo detector data Lo1, Lo2, ..., Lo of the object being examined. n ,...,Lo N The Lo detector element detects the corresponding real Hi detector data Hi1, Hi2, ..., Hi of the inspected object. n Hi N , where 1≤n≤N, and N are the number of Hi detector elements and / or Lo detector elements in the row direction of the dual-energy X-ray detector.

[0077] The actual Hi and Lo detector elements of the dual-energy X-ray detector are named Hi1, Lo1, Hi2, Lo2, ..., Hi n Lo n Hi N Lo N The sequences are read out in pairs, with multiple readout sequences forming a two-dimensional Hi / Lo matrix. Depending on how the offset is actually achieved relative to the readout direction of the detector elements, the above sequence can also start with element Lo1 and then proceed with element Hi. N Finish.

[0078] The virtual Hi- and / or Lo- detector data can be computed using an adapted demosaic algorithm (such as the one described in Li / Randhawa).

[0079] In order to properly adjust the demosaic algorithm (e.g., the algorithm described in Li / Randhawa), the Bayer pattern with three colors based on the algorithm described there is initially reduced to a two-color chessboard pattern, and one color of the chessboard is assigned to the Hi detector data, while the other color of the chessboard is assigned to the Lo detector data.

[0080] Then, the demosaic algorithm adapted to the chessboard pattern is applied accordingly to a two-dimensional Hi- / Lo matrix with the acquired real Hi- and Lo detector data rotated by 45° to calculate the virtual detector data associated with each real detector data.

[0081] The fourth aspect of this disclosure relates to a processing apparatus for processing Lo detector data Lo1, Lo2, Lo3, ..., Lo provided by the X-ray inspection apparatus of the second aspect. N and Hi detector data Hi1, Hi2, Hi3, ..., Hi N The processing device is configured to perform the method of the third aspect.

[0082] The fifth aspect of this disclosure relates to a system comprising an X-ray inspection apparatus of the second aspect and a processing apparatus of the fourth aspect, wherein the X-ray inspection apparatus is configured to provide Hi detector data and / or Lo detector data to the processing apparatus based on the object being scanned, and is connected to the processing apparatus for corresponding data communication with the processing apparatus.

[0083] The sixth aspect of this disclosure relates to a computer program product comprising software means for implementing the method of the third aspect when a computer program is implemented on a computer, such as the processing means according to the fourth aspect. Specifically, a computer program product comprising instructions that, when the computer program is implemented by a computer, particularly the processing means (300) according to the fourth aspect, cause the computer program to perform the method according to the third aspect.

[0084] The seventh aspect of this disclosure relates to a computer-readable medium that includes the computer program product of the sixth aspect.

[0085] The eighth aspect of this disclosure relates to a data stream comprising an electronically readable control signal interactive with a programmable computer, such that when the computer executes the electronically readable control signal, the computer performs the process of the third aspect. That is, transmitting a data carrier signal of the computer program product of the sixth aspect. Attached Figure Description

[0086] Further advantages, features, and details of the solutions (one or more) proposed herein will become clear from the following description, in which embodiments are described in detail with reference to the accompanying drawings. In this regard, features mentioned in the claims and specification may be necessary individually or in any combination. Similarly, the foregoing features and the features further described in detail herein may be used individually or in any combination. Functionally similar or identical parts or components partially bear the same reference numerals. The terms “left,” “right,” “top,” and “bottom” used in the description of the embodiments refer to the orientation of the drawings having generally readable reference numerals or symbols. The illustrated and described embodiments should not be construed as definitive, but rather have exemplary features for explaining the solutions proposed herein. The detailed description is intended for information to those skilled in the art, and therefore known structures and processes are not shown or explained in detail in the description so as not to complicate the understanding of the description.

[0087] Figure 1a This is a simplified perspective view of the structure of a conventional dual-energy detector.

[0088] Figure 1b yes Figure 1a A cross-sectional view of the detector through the xy plane.

[0089] Figure 2a This is a simplified perspective view of the structure of a first embodiment of a first implementation of a dual-energy X-ray detector proposed herein.

[0090] Figure 2b yes Figure 2a A cross-sectional view of the detector through the xy plane.

[0091] Figure 2c This is a side view of a detector having spaced-apart rows of first and second detectors according to a first embodiment.

[0092] Figure 3 This is a simplified perspective view of the structure of a second embodiment of the second implementation of the dual-energy X-ray detector proposed in this paper.

[0093] Figure 4 This is a simplified perspective view of the structure of a third embodiment based on a combination of the first and second embodiments of the dual-energy X-ray detector proposed herein.

[0094] Figure 5 This is a side view of a tilted detector according to an embodiment of the third implementation, wherein, as in Figure 2c Effective offset is achieved as shown in the embodiment according to the first embodiment.

[0095] Figure 6 It uses the dual-energy X-ray detector proposed here (such as...) Figure 2a-5 A simplified side view of an X-ray inspection device (the detector).

[0096] Figure 7 The following is shown for reference. Figure 1a and 1b Detector data readout status ( Figure 7 (top) and according to the first embodiment Figure 2a and 2b (or 5) detector data readout status Figure 7 bottom).

[0097] Figure 8 It shows the use of Figure 1a and 1b The detector is used as a reference ( Figure 8 (top) and utilization Figure 3 The detector utilizes a dual-energy X-ray detector according to the second embodiment. Figure 8 The detector data readout status at the bottom.

[0098] Figure 9 It shows the use of Figure 1a and 1b The detector data readout status as a reference detector ( Figure 9 (top), and how to use the detector in Figure 1 to achieve a similar effect through modified readout sequences of Hi- and / or Lo- detector elements. Figure 8 The readout status shown ( Figure 9 bottom).

[0099] Figure 10 It shows the use of Figure 1a and1b The detector data readout status as a reference detector ( Figure 10 (top) and utilization Figure 3 The readout of detector data used as a second reference detector ( Figure 10 (Central) and utilization Figure 3 The pixel offset of the detector in the z-direction corresponds to Figure 8 The readout of detector data at twice the pixel offset ( Figure 10 bottom).

[0100] Figure 11-13 An embodiment of a method for processing detector data acquired using the dual-energy X-ray detector proposed in this paper based on a demosaicing algorithm is shown.

[0101] Figure 14-17 The process for processing and preparation for use is shown. Figure 1a-5 The method shown illustrates a dual-energy X-ray detector that acquires real detector data to provide additional virtual detector data as needed. Specific Implementation

[0102] Dual-energy X-ray radiography is based on the fact that the attenuation coefficient is a property of the material through which radiation passes. An initial grayscale image is obtained by scanning the object with X-rays, where the gray level of a pixel corresponds to the corresponding measured intensity of the X-ray. By comparing the intensity of the X-rays used with the intensity measured at the image points, the attenuation coefficient of the material through which it passes can be determined. This attenuation coefficient is a function of the nuclear charge number, the density of the irradiated material, and the original energy of the X-rays used. If, for example, the object is scanned twice spectrally selectively using X-ray energies with different spectral frequencies, the dependence on material density can be eliminated from the corresponding equation.

[0103] To perform the two required spectrally selective measurements simultaneously, a so-called dual-energy detector has been developed, which has a structure that essentially has two spectrally selective energy channels. Therefore, this detector simultaneously provides Hi(gh) detector data induced by high-energy X-rays and Lo(w) detector data induced by low-energy X-rays. Using the difference in magnitude between the Lo- and Hi- detector data, the nuclear charge number (also the atomic number, proton number) of the material being irradiated, or the effective nuclear charge number of the material being irradiated in the object being examined, can be inferred for each image point (pixel) corresponding to the detector element.

[0104] Organic materials have approximately 10 or fewer atoms, while metallic materials have more than 10 atoms. Therefore, materials within an inspected object can be classified based on the determined effective number of atoms, and pixels in the displayed X-ray image can be colored accordingly for optical material identification. This principle is used, for example, in the visual evaluation of X-ray images of baggage at airport checkpoints, where the inspection device colors image pixels in the displayed X-ray image by orange for predominantly organic compounds, blue for predominantly metallic materials, and green for mixtures of organic and metallic materials, enabling operators to identify optical materials.

[0105] exist Figures 1a to 6 Before the following description, it should be noted that the xyz coordinate system is indicated in the figures for orientation and mutual reference. According to the corresponding xyz coordinate system, the longitudinal direction of the detector row always extends in the x-direction, while the direction RX (simplified as an arrow bundle) of the X-rays to be detected incident on the detector element extends in the y-direction, and when using detector rows, the direction corresponding to the scanning direction extends orthogonally to the detector row in the z-direction. The scanning direction typically corresponds to the direction TD (e.g., the direction of transport of the object through the detector row and through the X-ray inspection apparatus) of transmission (e.g., ...). Figure 6 (Simplified version). That is, the longitudinal direction (x-direction) of the detector proposed herein is typically arranged transversely to the scanning direction in use (z-direction).

[0106] It should be noted that the representation of the detector used here is simplified compared to the U-shaped or L-shaped detector rows commonly used in practice. In the case of U-shaped or L-shaped detector rows and the fan-shaped X-ray fan aligned with them, the detector rows do not extend specifically in the transport plane of the object being examined. However, the legs of the detector rows are always orthogonal to the X-rays being detected, and also orthogonal to the transport or scanning direction z. Therefore, the considerations made here for I-shaped detector rows extending only in the xz plane can be readily transferred to U-shaped and L-shaped detector rows used in practice.

[0107] Figure 1a and 1b The structure of a conventional dual-energy X-ray detector 1 (hereinafter referred to as detector 1) is shown in cross-sectional form of detector row 2. Figure 1a This is a simplified perspective view of detector 1, to illustrate the structure. Figure 1b yes Figure 1a The projection of detector 1 onto the xy plane.

[0108] Detector row 2 consists of dual-energy detector elements 3 arranged side-by-side; for clarity, only four such elements are shown, although in practice the number is typically much higher (e.g., 1,000). For this purpose, each detector element 3 consists of a low-energy Lo(w) detector element 4 and a high-energy Hi(gh) detector element 5, sandwiched between the X-rays RX to be detected, with a filter layer 6 (e.g., made of copper) between them. During scanning of the object being inspected, the Lo detector element 4 generates Lo detector data primarily caused by the low-energy X-rays RX, while the Hi detector element 5 generates Hi detector data caused by the high-energy X-rays RX. Therefore, detector 1 has at least two output channels: one providing Hi detector data and one providing Lo detector data.

[0109] The effective area A4 of Lo detector element 4 and the effective area A5 of Hi detector element 5 are essentially the same size. The effective areas of Lo detector element 4 and Hi detector element 5 have a length L along the row direction (i.e., the longitudinal direction of detector row 2) and a width W orthogonal to it.

[0110] For example, the effective area of ​​a detector element can be 0.8 mm (length L in the row direction) multiplied by 0.8 mm (width W across the row direction). For instance, a typical detector row can be 80 cm in total in the row direction, making the row consist of approximately 1,000 detector elements. In principle, the size of the detector elements is within a technically reasonable range, and the length of the detector row is freely selectable.

[0111] In use, detector row 2 is typically arranged transversely to the transport direction TD of the object being inspected, allowing the object to be scanned line by line with X-rays RX. Detector 1 can, in principle, consist of several detector rows 2. That is, for Figure 1a and 1b The detector 1 can simultaneously acquire low-energy X-ray images based on transmission of low-energy X-rays and high-energy X-ray images based on transmission of high-energy X-rays by scanning the object line by line.

[0112] Figure 1b The actual Hi detector data Hi1, Hi2, Hi3, ..., Hi provided by the individual detector elements 4 and 5 are also shown. N And Lo detector data Lo1, Lo2, Lo3, ..., Lo NTo describe in detail the processing of the real detector data and the associated readout process described below, the real readout Hi and / or Lo detector data is indexed along detector row 2: Hi or Lo classifies the corresponding Hi detector data or Lo detector data, which is indexed by position variables n = 1, 2, ..., N for the corresponding positions of the relevant detector elements in detector row 2, where 1 ≤ n ≤ N, and N is the number of detector elements of the relevant type (Hi or Lo) in the row direction. Figure 1b In a highly simplified example, N=8, meaning detector row 2 consists of 8 Hi detector elements and Lo detector elements arranged along the detector row.

[0113] exist Figure 6-8 The left part shows the typical readouts of the Hi- and / or Lo- detector data acquired by detector 1, showing the changes in position variable n along the direction of travel of the detector row and with time t.

[0114] Figure 2a-4 Each illustrated shows a specific embodiment of the dual-energy X-ray detector 100 presented herein, each essentially consisting of a first detector row 110 having a first detector element 111 and a second detector row 120 having a second detector element 122 arranged parallel thereto. In all embodiments, the two detector rows 110 and 120 are configured to have different spectrally selective responses to the spectrum of the X-ray RX to be detected. For this purpose, a corresponding first detector element 111 in the form of a Lo detector element for primarily detecting low-energy X-rays is arranged side by side in the first detector row 110, and a corresponding second detector element 122 in the form of a Hi detector element for primarily detecting high-energy X-rays is arranged side by side in the second detector row 120. In principle, according to the improvement presented herein, the two detector rows 110 and 120 are arranged parallel to each other in the row direction and one after the other in the direction of the X-ray RX to be detected, such that the detector rows 110 and 120 are offset from each other in an overlapping manner.

[0115] In the illustrated embodiment, each Hi detector element and each Lo detector element have the same length L in the row direction (i.e., the longitudinal direction of detector 100) and the same width W in a direction orthogonal to it. In the illustrated embodiment, the length L Hi and L Lo and width W Hi and W Lo Equal, that is, the effective areas of Hi detector element 111 and Lo detector element 122 are equal.

[0116] Figure 2aA simplified perspective view of a first embodiment of a first implementation of the dual-energy X-ray detector 100 proposed herein is shown to illustrate the basic construction. Figure 2b Through such Figure 2a A cross-sectional view of the detector 100 in the xy plane shown in the middle section.

[0117] exist Figure 2a and 2b In the diagram, the first detector row 110 and the second detector row 120 are offset from each other by a first offset Δx in the row direction. In the illustrated embodiment, the offset Δx corresponds precisely to half the length L, i.e., Δx = L / 2.

[0118] Figure 2b Show (similar to) Figure 1b The true Hi detector data Hi1, Hi2, Hi3, ..., Hi provided by individual detector elements 111 and 122 N And Lo detector data Lo1, Lo2, Lo3,....Lo N The following description details the processing of real Hi detector data and the related readout process. Figure 1b As shown, the actual readout Hi and / or Lo detector data is indexed along the detector rows. Hi or Lo categorizes the corresponding Hi detector data or Lo detector data, which is indexed by position variables n = 1, 2, ..., N. These position variables n = 1, 2, ..., N represent the corresponding positions of the relevant detector elements in detector row 2, where 1 ≤ n ≤ N, and N is the number of detector elements of the relevant type (Hi or Lo) along the row direction. Figure 2b In the highly simplified representation, N=8, that is, the detector row consists of 8 Hi detector elements and Lo detector elements arranged along the detector row.

[0119] In the first detector row 110, a Lo detector element 115 is shown that provides Lo detector data Lo4. However, with Figure 1a and 1b Unlike detector row 1, Lo detector element 115 does not have an associated Hi detector element. To address this issue, it is proposed to calculate the missing Hi detector data in Lo detector data Lo4 in the form of virtual detector data vHi4, based on existing Hi- and / or Lo- detector data.

[0120] As described above, the inventors discovered that calculating the virtual Hi detector data vHi... # and Lo detector data vLo #Various methods are possible, and they can be used individually or in combination.

[0121] For example, in a particularly simple method, the virtual Lo detector data vLo can be calculated based on the average or median of neighboring Hi and / or Lo detector data. # Or Hi detector data vHi # For example, the missing Hi detector data from the real detector data Lo4 obtained and provided by Lo detector element 113 can be calculated as the virtual Hi detector data vHi4 as follows:

[0122] Based on two directly adjacent real Hi detector data Hi4 and Hi5 as a certain first number (i.e., the detector data of two Hi detector elements 123 and 127 that overlap with the Lo detector element 115 under consideration, i.e., those adjacent to it on the left and right in the row) and three nearest real Lo detector data Lo3, Lo4 and Lo5 as a certain second number (i.e., the detector data of the Lo detector element 115 under consideration and the detector data of two Lo detector elements 113 and 117 that are adjacent to it on the left and right), the virtual Hi detector data vHi4 can be calculated.

[0123] Therefore, when calculating the virtual Hi detector data vHi4, the behavior of the Lo detector data values ​​can be considered through the values ​​Lo3, Lo4, and Lo5. The values ​​of Hi3 and Hi4 can be averaged and used as the base value of the virtual Hi detector data vHi4. This method can also be used as the basis for machine learning algorithms (e.g., deep learning algorithms).

[0124] As a result, using Figure 2a and 2b The dual-energy X-ray detector 100 of the embodiment can achieve double the spatial resolution in the X-ray image derived from detector data in the row direction of the detector 100.

[0125] Figure 2c This is a side view of the detector 100 according to the first embodiment, wherein a first detector row 110 having Lo detector elements 111 and a second detector row 120 having Hi detector elements 122 are arranged at a distance D (e.g., D = 1 cm) from each other. The detector rows 110 and 120 are offset from each other, overlapping effectively by an offset Δx in the row direction. X-rays RX incident on the detector 100 and passing through the detector elements are substantially orthogonal to the first and second detector rows 110 and 120.

[0126] exist Figure 3The image shows an embodiment of a second implementation of the dual-energy X-ray detector 100 proposed herein. The first detector row 110 and the second detector row 120 are now offset from each other by a second offset Δy orthogonal to the row direction. In the illustrated embodiment, the offset Δy corresponds precisely to half the width W, i.e., Δy = W / 2.

[0127] To calculate the missing detector data, it is possible to... Figure 2a and 2b Similar to the explanation in [the original text], through each real detector data source Hi # or Lo # Calculate the relevant virtual detector data vLo # or vHi # To process the provided detector data.

[0128] result, Figure 3 The dual-energy X-ray detector 100 of the embodiment can be used to increase the spatial resolution of the X-ray image derived from detector data in the scanning direction and / or increase the band velocity, taking into account the ratio of scanning velocity or band velocity.

[0129] Figure 4 A first embodiment of the dual-energy X-ray detector 100 proposed herein is shown. Figure 2a and 2b ) and second implementation method ( Figure 3 An embodiment of the combination of ) is described. Specifically, the first detector row 110 and the second detector row 120 are offset from each other in the row direction by a first offset amount Δx, where the offset Δx corresponds to half the length L, i.e., Δx = L / 2. The first detector row 110 and the second detector row 120 are also offset from each other orthogonally in the row direction by a second offset amount Δy, where in the illustrated embodiment, the offset Δy corresponds to exactly half the width W, i.e., Δy = W / 2. As elsewhere stated, L = W. Therefore, utilizing Figure 4 In this embodiment, it is possible to simultaneously achieve... Figure 2a-3 The examples in b illustrate the effects of the particularly preferred first and second embodiments.

[0130] Figure 5 This is a side view of the conventional detector 1 in Figure 1, serving as an example of using a conventional detector to obtain an effective offset by means of an arrangement according to the third embodiment of this disclosure. Strictly compared to Figure 1, the difference lies in that the Lo detector element 4 and Hi detector element 5 of detector row 2 of detector 1 are spaced apart by a distance D (e.g., D = 1 cm); this allows for... Figure 2c Good comparability.

[0131] exist Figure 5In detector 1, there is no local offset between Lo detector element 4 and Hi detector element 5, wherein the Lo detector elements and Hi detector elements are arranged in a 1:1 ratio with each other. However, by means of a third embodiment of this disclosure, the desired effective offset between Lo detector element 4 and Hi detector element 5 can be achieved by tilting detector row 2 at a certain angle relative to the incident X-ray RX, for example, similar to Figure 2c The offset in detector 100. Figure 5 This is illustrated in a very simplified manner by showing how, through this tilted arrangement of a conventional detector row 2 relative to the incident X-ray RX, the desired effective deflection can be achieved, for example, in... Figure 2c The effective offset is implemented in the first embodiment.

[0132] For better understanding, the normal N passing through the center of the two lines formed by the relevant Hi and Lo detector elements is plotted on... Figure 5 Similarly, a reference X-ray beam RXref incident on detector 1 is also plotted, which is the point through which the normal N of detector 1 is intersected. Therefore, a tilt angle is formed between the normal N and the reference X-ray beam RXref.

[0133] exist Figure 5 In this process, the tilt angle is adjusted to provide the desired effective offset Δx between the Lo and Hi detector elements 4 and 5. Therefore, in Figure 5 In the example, it actually achieves the same as Figure 2c The first embodiment has the same offset as the corresponding embodiment.

[0134] Therefore, in Figure 5 The principle of the third embodiment shown can also be adapted to the second embodiment. Similarly, the third embodiment can be combined with the first and / or second embodiments. Finally, by adjusting the detector accordingly, the combined effects of the first and second embodiments can also be achieved solely through the third embodiment.

[0135] Figure 6 A system 400 is shown, which is basically composed of an X-ray inspection device 200 and a processing device 300.

[0136] The X-ray inspection apparatus 200, shown in a highly simplified form, has radiation shielding curtains 202 and 204, one of which is arranged at each of the input 206 and output 208 of the radiation tunnel 210 of the X-ray inspection apparatus 200. Between the radiation shielding curtains 202 and 204, within the radiation tunnel 210, is a radiation area 212 having at least one radiation source 214 (e.g., an X-ray tube) and a collimator for generating an X-ray fan 215 aligned with the dual-energy X-ray detector 100 presented herein. A conveying device 218 (e.g., a sliding belt conveyor) is used to convey baggage items 216, which are to be inspected, through the radiation tunnel 210 in a conveying direction TD. The linear detector 100 is L-shaped or U-shaped and arranged such that its longitudinal direction is orthogonal to the conveying direction TD, such that the conveying direction TD corresponds to the scanning direction of the item 216 to be inspected.

[0137] The processing device 300 is substantially configured to perform at least one of the methods proposed herein for processing Hi detector data and / or Lo detector data acquired by the dual-energy X-ray detector proposed herein. For example, the dual-energy X-ray detector 100 is as follows: Figure 2a-5 The detector is shown in a simplified form. Detector data provided by the dual-energy X-ray detector 100 and processed by the processing device 300 can be used to generate an X-ray image of the object 216 under inspection, which is colored based on the material category and displayed to the operator on a screen (not shown) in a manner known per se.

[0138] The processing device 300 may be part of the control device 220 of the X-ray inspection device 100, such as... Figure 6 As shown. The processing unit 300 can, in principle, be completely separate from the X-ray inspection unit 200, for example, in a central location where the raw detector data from several inspection units 200 converge and are centrally processed. This is no different from the proposed measures for processing detector data.

[0139] The processing unit 300 may also be part of the detector 100, such that the detector data generated by the detector 100 has already been processed at the detector 100 according to the measures proposed herein. Therefore, the detector 100 proposed herein is, in principle, compatible with existing X-ray inspection apparatuses having conventional detector units. That is, as a result, the proposed implementation of the new detector 100 with integrated processing of detector data can be used in X-ray inspection apparatuses that are structurally identical, with lower system cost and constant image quality. Alternatively, the spatial resolution of existing X-ray inspection apparatuses can be increased at nearly the same system cost.

[0140] Now for reference Figure 2b and7 -10 illustrates a particularly preferred method for processing Hi detector data and / or Lo detector data provided by the dual-energy X-ray detector 100 proposed herein. In principle, various methods are conceivable for calculating virtual detector data for detector elements lost due to offset. The basic idea behind the improvement presented here lies in the overlapping offset Hi and Lo detector elements of the dual-energy X-ray detector 100, such as... Figure 2a-5 As shown in the image.

[0141] Reference Figure 2b First, the task of processing the provided real Hi and / or Lo detector data is explained.

[0142] Figure 2a The detector setup shown with overlapping Hi- and Lo-detector rows 110 and 120 leads to the problem that not every real Lo-detector data (e.g., Lo-detector data Lo4) can be accurately assigned to a Hi-detector data, and vice versa. Therefore, another type of corresponding detector data associated with a particular real Hi-detector data (Hi or Lo) must be supplemented with virtual detector data appropriately determined through calculation or otherwise. Figure 2b In the diagram, the virtual detector data vHi4 is depicted as an example used for the real detector data Lo4.

[0143] "Real detector data" is understood here as the measured intensity value of X-rays actually detected using a specific detector element physically present at detector 100.

[0144] "Virtual detector data" is understood here to represent virtual intensity values ​​generated at specific locations of detector elements that do not physically exist in detector 100 through calculation or other means.

[0145] If the value of a specific Lo detector element 111 (e.g., detector data Lo4) takes the value x, then the associated Hi detector data can take values ​​from x to x+a, depending on the material being transmitted. For plastics, the difference a will be very small, i.e., the sought Hi detector data vHi4 and Lo detector data Lo4 will be approximately the same. For steel, the difference obtained in a will be significant. It is precisely this situation—as described above—that is used for material detection by dual-energy X-ray irradiation. That is, to distinguish steel from plastic or aluminum at a certain location, information from the Hi and Lo channels of the dual-energy X-ray detector 100 is needed for that location. If the value of the associated Lo detector data can be directly inferred from the Hi detector data, then two different spectral selectivity measurements are not required.

[0146] The inventors have discovered that, instead of real Lo or Hi detector data, relevant missing values ​​can be identified using virtual vLo or vHi detector data with sufficient accuracy. In this case, for practicality, it is desirable to achieve results as close to reality as possible with minimal computational power.

[0147] A simple approach is, for example, to calculate the missing real Hi detector data for the real Lo detector data Lo4 as the average of two adjacent real Hi detector data sets Hi4 and Hi5. However, this simple approach does not achieve the required accuracy.

[0148] The inventors also discovered that the behavior of the Hi detector data in the Hi channel of detector 100 is similar to that of the Lo detector data in the Lo channel. In other words, when the signal in the Hi channel decreases, under real conditions, it also decreases in the Lo channel, and vice versa. This depends on the material through which the radiation passes, which is more or less intense. That is, the behavior of the detector data in the Lo channel, i.e., in Lo detector row 110, can be used to determine the virtual Hi detector value vHi4. For example, as described elsewhere, a deep learning algorithm trained to compute the missing virtual Hi and / or Lo detector data can be used.

[0149] Before describing in detail the particularly preferred method, which is surprisingly feasible and requires low computational power to provide practical results, as shown here, refer to... Figure 7-10 This illustrates how various possible configurations of the dual-energy X-ray detector 100 presented herein (such as...) Figure 2a-5 As shown, the desired improvement in spatial resolution is achieved in the generated X-ray image.

[0150] exist Figure 7-10 In each case in the left part of the figure, the timing of the readout process on a highly simplified dual-energy X-ray detector 100 (hereinafter referred to as detector 100) is shown, for example as follows: Figure 2a As shown, each of its Hi- and Lo- detector rows has only 4 associated real detector elements. Real Hi detector elements are shown as empty / white circles, and real Lo detector elements are shown as solid / black circles.

[0151] refer to Figure 6The X-ray inspection apparatus 200 shown as an example briefly describes the provision of real Hi and / or Lo detector data. Detector readouts of the detector 100 are performed sequentially. The real detector data output by each Hi and / or Lo detector element is indexed by the Hi or Lo type, where the first digit after Hi or Lo is the position variable n in the detector row, and the second digit is the sequential number of the detector readout that has occurred. For example, Lo... 32 This is the Lo detector data of the 3rd (n=3) Lo-detector element in the 2nd (t=2) readout.

[0152] For example, such as Figure 6 As shown, the conveyor belt 218 moves continuously, thereby continuously conveying the inspection object 216 through the X-ray fan 215 in the conveying direction TD. Due to the continuous movement of the inspection object 216 relative to the detector 100 and the time-sequential readout of each detector element, the X-ray image quality is significantly improved. Figure 7-10 The corresponding readouts shown on the left are the actual Hi and / or Lo detector data.

[0153] In respectively Figure 7-9 The upper part and Figure 10 The upper third of the section shows Figure 1a and 1b The readout of conventional detector 1 is shown here. Here, the Hi detector data of the first Hi detector element is read out. 11 Then, the Lo detector data of the first Lo detector element is read out. 11 Then read the Hi detector data of the second Hi detector element. 21 Then, the Lo detector data of the second Lo detector element is read out. 21 ; etc. Once the Lo detector data of the last Lo detector element has been read... n1 The process then restarts at the first Hi detector element, where the Hi detector data is read out. 12 .

[0154] Figure 7-10 The right-hand side shows a two-dimensional matrix M#, whose fields are assigned to the acquired real Hi- and / or Lo- detector data. Each field of matrix M# with associated detector data corresponds to the corresponding pixel in the X-ray image to be derived from the detector data. Therefore, it is possible to derive the X-ray image from the corresponding... Figure 6-9 The effect of the proposed modifications to the relevant detector 100 on the spatial resolution of the X-ray image derived from the relevant detector data can be immediately seen in matrix M#. Figure 1a and 1bThe upper right half of the matrix M0 provided by the conventional detector 1 is used as the... Figure 7-10 The reference for matrices M6-M10.

[0155] During scanning, the continuous movement of the object under examination and the sequential readout of the actual detector elements effectively result in a temporal offset between two adjacent readout processes and a spatial offset within the pixels of the X-ray image assigned to the individual detector data. Figure 7-10 The effective offset is no longer shown in the right part of the matrices M0 and M6-M10, but it will still be visible in the associated X-ray images if it is not compensated again by shifting the individual image rows relative to each other by default.

[0156] Now, referring to Figure 7 The upper part, targeting according to Figure 2a and 2a The detector 100 of the first embodiment shows the readout condition. On detector 100, the Hi detector row is offset from the Lo detector row in the row direction and perpendicular to the transport direction TD. Figure 6 The deviation Δx corresponds to half the length L of the detector element. In the same order as detector 1 ( Figure 7 The upper half of the data reads out the actual Hi- and / or Lo- detector data. Figure 7 The readout results show that not every real Hi detector data can be assigned to a low detector data, and vice versa.

[0157] Reference Figure 7 Compared with matrix M0 in the same figure, matrix M6 shows that the resolution in the detector row direction (i.e., perpendicular to the band running direction or scanning direction) is improved, i.e. doubled.

[0158] If the spatial resolution of the X-ray image derived from the detector data increases in the row direction of detector 100 (i.e., orthogonal to the transport or scan direction), then the configuration of detector row 100 and Figure 7 The readout characteristics are of great interest in practice. The resolution of X-ray images in this dimension cannot be achieved simply by increasing the readout frequency. The only option is to use more and smaller detector elements, which will increase the cost per detector and increase noise in the detector data, thus deteriorating the signal-to-noise ratio.

[0159] Figure 8 It shows according to Figure 3 The second embodiment of the detector 100 readout status, that is, on the detector 100, the Hi detector row 120 and the Lo detector row 110 are orthogonal to the row direction (i.e., in the transmission direction TD (reference)). Figure 5The offsets are equal to the offsets on top, overlapping each other, and the offset is a Δy corresponding to half the width W of the detector element. If we use the offsets on top, the offsets are equal to the offsets on top, overlapping each other, and the offset is equal to the offset Δy corresponding to half the width W of the detector element. Figure 7 If the real Hi-detector and Lo-detector elements are read out in the same order, then the following is obtained: Figure 8 The readout status is shown in the figure.

[0160] refer to Figure 8 Matrix M7, and matrix M0 of the same graph (similar to...) Figure 7 The readout situation is also shown in Figure 7 In the readout scenario, it can be seen that there is no longer a specific Lo detector data that can be explicitly assigned to each Hi detector data. However, Figure 8 Matrix M7 shows the scanning direction (or...) Figure 6 The spatial resolution in the X-ray image derived from the detector data is improved by doubling the transmission direction (TD).

[0161] If the obtained X-ray images are in the scanning direction or the transmission direction (TD, Figure 6 The spatial resolution of the X-ray machine can be increased, or the transmission speed of the object being inspected can be increased while maintaining the same resolution by appropriately adjusting the ratio of readout frequency and belt speed. Figure 8 The reading results are interesting.

[0162] For the overlap offset Δy between Hi-detector row 120 and Lo-detector row 110 in the scan direction (orthogonal to the row direction of detector 100), it has been shown that it is particularly suitable to define the offset Δz as (mb) / (2f), where b is the band velocity in cm / s, f is the readout frequency in 1 / s, and m is an odd integer (m = 1, 3, 5, 7, ...). Alternatively or supplementarily, spatial resolution can be increased in the scan direction or in a direction orthogonal to the row direction of the detector by increasing the readout frequency f.

[0163] Now, for reference Figure 9 This reveals another ingenious consideration by the inventor. The inventor has realized that, by means of... Figure 1a and 1b Modified readout patterns (i.e., readout sequences) of Hi- and / or Lo- detector data on a conventional dual-energy X-ray detector 1, as implemented in a quasi-hardware manner using the second embodiment of detector 100, can also achieve similar effects at the achieved spatial resolution (or alternatively, the transmission or scan rate to be set) in a conventional detector 1 (such as...). Figure 1a and 1b This is achieved in the X-ray inspection device shown.

[0164] When using a dual-energy X-ray detector row with one high-energy Hi detector element and one low-energy Lo detector element per pixel, the following may occur: Figure 9 The readout configuration shown in the diagram is such that the two high-energy Hi detector elements and the low-energy Lo detector elements are arranged substantially in the same direction as the X-rays to be detected, one above the other or one behind the other.

[0165] like Figure 9 As shown on the left, the readout sequence at detector 100 can be modified such that initially all detector elements of one type (Hi or Lo) are read along the position variable n of the detector row, and subsequently all detector elements of another type are read along the position variable n of the detector row. That is, for example, as... Figure 9 As shown, initially all Hi detector data Hi1, Hi2, Hi3, ... Hi are read out. N Then read out all Lo detector data: Lo1, Lo2, Lo3, ..., Lo N (vice versa).

[0166] A comparison of matrices M7 and M8 shows that, using Figure 9 The modified readout sequence can achieve the same result as that achieved in the detector 100 according to the second embodiment by shifting the first detector row 110 and the second detector row 120 relative to each other. Therefore, Figure 9 The reading method is Figure 3 and 8 The equivalent software solution for the hardware solution is where the Hi-detector and Lo-detector elements overlap each other. In other words, if the Hi-detector and Lo-detector rows 110, 120 on the detector are not offset from each other in the scanning direction, but the readout order of the Hi-detector and Lo-detector elements is appropriately changed, it is possible to achieve the same software solution as the hardware solution. Figure 3 and 8 The same effect on spatial resolution applies to the detector row 100. As a result, spatial resolution is also improved in both the transport and scanning directions in this case.

[0167] Figure 9 The readout performance is excellent because virtually no modifications are needed to the conventional detector 1 (see [reference]). Figure 1a and 1b Modifications may be needed. To implement the modified timing readout sequence, the circuit board of detector 1 or the readout chip may need to be redesigned. If the readout chip interconnected with detector 1 can internally change the timing of the readouts of the individual real Hi and Lo detector elements as needed, then this can be achieved. Figure 9 The readout scenario requires no major measures.

[0168] Now refer to Figure 10 This illustrates another advantage of the second implementation. For a given pair of values ​​(b1, f1), the offset Δz = (mb) / (2f) can be calculated, which leads to... Figure 10 The middle example illustrates a situation where the missing values ​​must be calculated.

[0169] Since X-ray machines are typically integrated into complex conveyor systems, it may sometimes be desirable to adapt the belt speed to these conditions.

[0170] The inventors have recognized that it is unnecessary to develop different types of devices to achieve multiple transmission speeds, because the second embodiment of this disclosure provides a high degree of flexibility. Now, referring to... Figure 9 This demonstrates another ingenious consideration of the inventor, which can be used as explained below.

[0171] Other pairs of values ​​(b2, f2) can belong to the same offset Δz, and the following occurs at these values. Figure 10 In the case shown, b2 = b1 / 2 or f2 = 2f1, which corresponds to half the belt speed or twice the readout frequency. Here, the forward movement of the baggage in the conveying direction precisely corresponds to the offset Δz, and for each measured Hi detector data, there is again a true measured Lo detector data, and vice versa. Therefore, the X-ray system can operate in two different modes with an offset Δz.

[0172] Now, referring to Figure 11-14 This paper explains an embodiment of a method for processing real Hi and / or Lo detector data acquired using the dual-energy X-ray detector 100 proposed herein.

[0173] To strike a balance between a good algorithm and an efficient algorithm for processing real Hi and / or Lo detector data, the inventors found that an algorithm from the field of digital optical photography was suitable. Digital photography uses image sensors with an upstream color filter array (CFA), such that each individual pixel of the image sensor can detect only one of the three primary colors (red, blue, or green). The most common pattern for this CFA is the Bayer pattern described above.

[0174] The inventors have recognized that the detector 100 proposed herein ( Figure 2a-6 (and 7-8) or by a modified readout method ( Figure 9 The structure of the matrix M6-M10 provided may be able to be processed in a manner similar to that of image data acquired by an image sensor with a Bayer CFA, relative to the Hi- and / or Lo- detector data contained therein.

[0175] In the case of image data generated by an image sensor with Bayer CFA, for each pixel, two additional color information must be supplemented through appropriate calculations. Figure 7-9 For each real detector data of one type (Hi or Lo), another type of virtual detector data must be calculated.

[0176] The inventors propose to modify the CFA demosaic used in digital image sensors for processing real detector data of the detector 100 presented herein. Since many known implementations of CFA demosaic exist, it should be sufficient to explain the basic principles employed herein. In fact, this disclosure is by no means limited to any particular single or specific implementation.

[0177] The core idea of ​​CFA desacrifice is to calculate the two missing colors for each pixel of the Bayer pattern, based on the known colors detected in the nearby environment.

[0178] Accordingly, this means that for matrix M6*, for example, the missing virtual Hi detector data 44 at the location of the real Hi detector data 44 can be calculated from the real Hi and / or Lo detector data obtained from the nearby environment.

[0179] Figure 11 First, the Bayer pattern (BAYER) is shown. Figure 11 Comparison of the left-hand side with the Hi- and / or Lo- detector data matrix M6* ( Figure 11 (on the right), the data matrix M6* is, for example, in... Figure 7 In the case of reading, use Figure 2a and 2b The detector generates this. In the Bayer pattern (BAYER) Figure 11 In the matrix M6* (left side), white boxes represent blue pixels, shaded boxes represent green pixels, and intersecting shaded boxes represent red pixels. Figure 11 (Right side), Lo-detector data is represented by a shaded box, and Hi-detector data is represented by a cross-shaded box.

[0180] exist Figure 12 middle, Figure 11 The Bayer pattern has been reduced to two colors. That is, the modified Bayer pattern is now a checkerboard, with the remaining two colors being red and green.

[0181] exist Figure 13 In, with Figure 12In contrast, simply rotating matrix BAYER* clockwise by 45° represents appropriately allocating the data of matrix M6* to the data of the modified Bayer pattern BAYER**. Figure 13 It is also shown that the modified Bayer pattern BAYER** and matrix M6* have a similar alternating structure. For example, it is possible to map pixels in the marked region of the modified Bayer pattern BAYER** to detector data in the marked region of matrix M6*.

[0182] Based on the above, the CFA demosaic algorithm for Bayer patterns can be applied to detector data in matrix M6*.

[0183] The inventors found that the results were slightly worse because the detector size was more widely spaced in both the horizontal and vertical directions than the pixels in the Bayer pattern. However, the vertical detector data sequence can be taken into account in the M6* matrix, which can compensate for this, whereas the vertical detector data sequence does not exist in the Bayer pattern.

[0184] The algorithm used to compute missing values ​​is based on the publication "Color filter arraydemosaicking using high-order interpolation techniques with a weighted median filter for sharp color edge preservation" by Li and Randhawa (IEEE Transactions on Image Processing, Vol. 18, No. 9, September 2009), the contents of which are incorporated herein by reference.

[0185] In this work, based on Taylor series expansion of nearby measurements, interpolation values ​​for missing pixels are first determined for each of a given number of spatial directions. Then, for each of these spatial directions, a gradient is calculated, which is a measure of how much the signal changes in the corresponding direction. With the help of these gradients, a weighted average is formed from all the interpolation values ​​in the final step.

[0186] Using the proposed demosaic method, the detector data matrix provided by the detector 100 yields surprisingly good results with unexpectedly low computational cost.

[0187] Finally, the following Figure 14-17 Various aspects of the method proposed herein for processing real Hi- and / or Lo- detector data obtained from an inspected object using one of the detectors 100 described herein are illustrated.

[0188] Figure 14 It shows a method for processing, for example Figure 6 The X-ray inspection device 200 provides Lo detector data Lo1, Lo2, Lo3, ..., Lo N and Hi detector data Hi1, Hi2, Hi3, ..., Hi N The basic structure of method 1300. Therefore, the method includes the following basic steps: Step S1, for calculating the actual Hi detector element 122 (e.g., see...). Figure 2b The corresponding virtual Lo detector data vLo at the location of the real Lo detector element 111 (e.g., see...) and step S2, for calculating the virtual Lo detector data vLo at the location of the real Lo detector element 111 (e.g., see...) Figure 2b The corresponding virtual Hi detector data vHi at the location of ).

[0189] In method 1300, step S1 for calculating virtual Hi detector data vLo at the location of the real Hi detector element 122 includes: in step S11, calculating the virtual Lo detector data based on a first number of real Lo detector data adjacent to the real Hi detector element 122 and a second number of real Lo detector data adjacent to the real Hi detector element 122.

[0190] In method 1300, step S2 for calculating virtual Hi detector data vHi at the location of the real Lo detector element 111 includes: in step S21, calculating virtual Hi detector data based on a determined first number of real Hi detector data adjacent to the real Lo detector element 111 and a determined second number of real Lo detector data adjacent to the real Lo detector element 111.

[0191] Step S11 includes step S12, which is used to calculate the virtual Lo detector data as the average value of the adjacent Hi and / or Lo detector data.

[0192] Step S21 includes step S22, which is used to calculate the virtual Hi detector data as the average value of the adjacent Hi and / or Lo detector data.

[0193] Figure 15 The modified method 1400 is shown, wherein step S1 includes step S14 of calculating virtual Hi detector data by considering the values ​​of Hi and / or Lo detector data, and step S2 includes step S24 of calculating virtual Hi detector data by considering the values ​​of Lo detector data.

[0194] exist Figure 14 and 15In methods 1300 and 1400 shown, a deep learning algorithm (DLA) can be used to implement the above calculation steps for virtual Hi and / or Lo detector data based on the provided Lo and Hi detector data.

[0195] Figure 16 It was shown as Figure 14 Method 1500 is a specific implementation of method 1300. As a prerequisite for method 1500, for example in... Figure 6 In the inspection system 200, detector data acquired and provided by the dual-energy X-ray detector 100 is indexed by a position variable n, which extends from one end in the row direction of the dual-energy X-ray detector 100, such that the Lo detector elements contain the actual Lo detector data Lo1, Lo2, ..., Lo of the inspected object. n ,...,Lo N Furthermore, the Lo detector element detects the corresponding real Hi detector data Hi1, Hi2, ..., Hi of the inspected object. n Hi N , where 1≤n≤N, and N are the number of Hi and / or Lo detector elements in the row direction of the dual-energy X-ray detector (100). Figure 15 Method 1500 has the following steps:

[0196] Step S31, using Hi1,Lo1,Hi2,Lo2,...,Hi n Lo n Hi N Lo H The actual Hi detector element 111 and Lo detector element 122 of the dual-energy X-ray detector 100 are read out in pairs in sequence (see... Figure 2b (Or vice versa, using Lo1, Hi1, Lo2, Hi2, ..., Lo) n Hi n ,...,Lo N Hi N (in order).

[0197] Step S32, using the sequences Hi1, Lo1, Hi2, Lo2, ..., Hi read in this way n Lo n Hi N Lo N (or Lo1, Hi1, Lo2, Hi2, ..., Lo) n Hi n ,...,Lo N Hi NThis forms a two-dimensional Hi / Lo detector data matrix M5-M10; M6* (see...) Figure 7-10 ).

[0198] Step S33 is used to calculate virtual Hi- and / or Lo- detector data using the adapted demosaic algorithm. The adaptation and application of the demosaic algorithm include:

[0199] Step S331 is used to reduce the three-color Bayer pattern based on the demosaic algorithm to a two-color checkerboard pattern Bayer*.

[0200] In step S332, the chessboard pattern BAYER* is calculated to rotate 45° clockwise.

[0201] In step S333, one color of the rotated chessboard pattern BAYER** is assigned to the Hi detector data and / or the Lo detector data, and another color of the chessboard is assigned to the Lo detector data.

[0202] Step S334: Apply the demosaic algorithm for the rotated checkerboard pattern BAYER** to the acquired two-dimensional Hi / Lo detector data matrix M6-M10; M6* is the real Hi- and / or Lo- detector data;

[0203] As a result, similar to how a photoelectric sensor chip with a BAYER color filter can calculate the other two primary colors missing for a given pixel, the required virtual (i.e., missing) detector data is calculated using an adapted demosaic algorithm.

[0204] Figure 17 It shows how to read out, such as Figure 1b The dual-energy X-ray detector 1 shown is configured with a readout method 1600 for the Hi detector element and the Lo detector element, and the dual-energy X-ray detector can be mounted, for example... Figure 6 In the X-ray inspection device 200. According to Figure 1b The dual-energy X-ray detector 1 has at least one dual-energy X-ray detector row 2, each pixel of which has a high-energy Hi detector element 5 and a low-energy Lo detector element 4, which are arranged vertically and horizontally in a substantially consistent manner in the direction of the X-ray to be detected RX. The Hi- and / or Lo- detector data to be read from detector row 2 is defined as Hi(n) or Lo(n) for the corresponding Hi- and / or Lo- detector data, with position variables n = 1, 2, 3, ..., N for the corresponding positions of the relevant detector elements in detector row 2, where 1 ≤ n ≤ N, and N is the corresponding number of Hi- and Lo- detector elements in the row direction.

[0205] Figure 17 The readout method 1600 basically includes step S4 for reading out Hi and Lo detector elements, wherein initially all detector elements of one type (i.e., Lo or Hi) are read out along the position variable n, and subsequently all other detector elements of the other type are read out along the position variable n.

Claims

1. A dual-energy X-ray detector (100) comprising a first detector row (110) having a first detector element (111) and a second detector row (120) arranged parallel thereto and having a second detector element (122), wherein, The detector rows (110, 120) are arranged parallel to each other in the row direction and one after another in the direction of the X-ray to be detected RX, such that the projections of the first detector row and the second detector row (110, 120) in the direction of one of the X-rays to be detected RX overlap each other with an effective offset Δx; Δz, wherein the X-ray passes through the surface centroid of the reference detector element of the first detector row or the second detector row (110, 120). The effective offset includes a first offset Δx in the row direction and a second offset Δz in the z direction, which is orthogonal to the row direction and the direction of the X-ray RX to be detected. The second offset Δz, orthogonal to the row direction, is defined as the ratio of the readout frequency f (in 1 / s) of the detector elements (111, 122) to the transmission speed b (in cm / s) of the object under inspection (216) relative to the dual-energy X-ray detector (100). , Where m is an odd integer, m = 1, 3, 5, 7, ...

2. The dual-energy X-ray detector (100) according to claim 1, wherein, The detector rows (110, 120) are configured to respond differentially to the spectra of X-rays RX to be detected by the first detector row (110) and the second detector row (120), the first detector row (110) having an associated first detector element (111) in the form of a Lo detector element to primarily detect low-energy X-rays, and the second detector row (120) having an associated second detector element (122) in the form of a Hi detector element to primarily detect high-energy X-rays.

3. The dual-energy X-ray detector (100) according to claim 1, wherein, The first detector row having the first detector element (4) and the second detector row having the second detector element (5) are spaced apart from each other by a predetermined distance D, and are tilted at an angle α relative to the reference X-ray beam RXref in the row direction and / or orthogonal to the row direction.

4. The dual-energy X-ray detector (100) according to claim 1, wherein... The first offset Δx in the row direction corresponds to half the width of the first and second detector elements (111, 122) in the row direction.

5. An X-ray inspection apparatus (200) comprising a dual-energy X-ray detector (100) according to any one of claims 1-4, wherein The X-ray inspection device (200) is configured to transport the object to be inspected (216) through the inspection device (200) in the transport direction TD; The row direction of the dual-energy X-ray detector (100) is arranged orthogonally to the transmission direction TD; and It is configured to provide first detector data Lo1, Lo2, Lo3, ..., Lo of the acquired inspected object (216). N Second detector data Hi1, Hi2, Hi3, ..., Hi N .

6. A method for processing first detector data Lo1, Lo2, Lo3, ..., Lo provided by the X-ray inspection apparatus (200) according to claim 5. N Second detector data Hi1, Hi2, Hi3, ..., Hi N The method, the method comprising: (S1) Calculate the corresponding virtual first detector data vLo at the location of the real second detector element (122); and / or (S2) Calculate the corresponding virtual second detector data vHi at the location of the real first detector element (111).

7. The method of claim 6, wherein The virtual first detector data vLo calculated at the location of the real second detector element (122) in (S1) includes: (S11) Based on a specific first number of real first detector data adjacent to the real second detector element (122) and a specific second number of real second detector data adjacent to the real second detector element (122), calculate the virtual first detector data; as well as The calculation of virtual high detector data vHi at the location of the real low detector element (111) in step (S2) includes: (S21) calculating the virtual second detector data based on a determined first number of real second detector data adjacent to the real first detector element (111) and a determined second number of real first detector data adjacent to the real first detector element (111).

8. The method of claim 6, wherein the method comprises at least one of the following steps: (S12) Calculate the virtual first detector data as the average of adjacent first detector data and second detector data and / or (S22) Calculate the virtual second detector data as the average of adjacent second detector data and first detector data; (S14) Calculate the virtual first detector data considering the behavior of the values ​​of the second detector data, and / or (S24) Calculate the virtual second detector data considering the behavior of the values ​​of the first detector data; and (S5) The virtual second or first detector data is calculated using a deep learning algorithm based on the provided first and second detector data.

9. The method according to claim 6, in, The detector data acquired and provided by the dual-energy X-ray detector (100) is indexed by a position variable n extending in the row direction of the dual-energy X-ray detector (100) starting at one end, such that the first detector element is the true first detector data Lo2, ..., Lo of the object being inspected. n , .... , Lo N The second detector element detects the corresponding real second detector data Hi1, Hi2, ..., Hi of the inspected object. n Hi N Wherein, 1 ≤ n ≤ N, and N are the number of the second detector element and the first detector element respectively in the row direction of the dual-energy X-ray detector (100); and the method further includes: (S31) Let Hi1, Lo1, Hi2, Lo2, ..., Hi n Lo n Hi N Lo N The actual second detector element and the first detector element (111, 122) of the dual-energy X-ray detector (100) are read out in pairs in sequence, or in Lo1, Hi1, Lo2, Hi2, ..., Lo n Hi n , ..., Lo N Hi N The actual second detector element and the first detector element are read out in reverse order; (S32) A two-dimensional detector data matrix M5-M10 is formed from the multiple sequences read out in the readout step; M6 ;as well as (S33) Using an adapted demosaic algorithm, calculate the virtual second and / or first detector data. This step includes: - (S331) Reduce the three-color Bayer pattern under the demosaic algorithm to a two-color checkerboard pattern Bayer. ; - (S332) Apply the two-color checkerboard pattern BAYER Rotate 45° clockwise; - (S333) rotate the chessboard pattern BAYER One color is assigned to the second detector data, and another color of the chessboard is assigned to the first detector data; and - (S334) Apply the appropriate chessboard pattern to the rotated BAYER The aforementioned demosaic algorithm is applied to the acquired two-dimensional detector data matrix M6-M10; M6 The actual data from the second detector and the data from the first detector.

10. A method for reading out first and second detector elements of a dual-energy X-ray detector (1), particularly in the X-ray inspection apparatus (200) of claim 5, wherein, The dual-energy X-ray detector (1) includes at least one dual-energy X-ray detector row (2), each pixel of the at least one dual-energy X-ray detector row having a high-energy second detector element (5) and a low-energy first detector element (4), the high-energy second detector element and the low-energy first detector element being arranged substantially aligned one on top of the other in the direction of the X-ray RX to be detected, wherein the second and first detector data of the detector row (2) to be read out are defined as Hi(n) or Lo(n) for the corresponding second or first detector data data, wherein the position variable n = 1, 2, 3, ... N, where 1 ≤ n ≤ N, and N is the corresponding number of second and first detector elements in the row direction, the readout method comprising: (S7) Read out the second and first detector elements such that initially all detector elements of one type are read out along the position variable n, and then all other detector elements of another type are read out along the position variable n.

11. A processing apparatus (300) for processing the first detector data Lo1, Lo2, Lo3, ..., Lo provided by the X-ray inspection apparatus (200) according to claim 5. N and the second detector data Hi1, Hi2, Hi3,..., Hi N ,in, The processing device (300) is configured to perform the method according to any one of claims 6-10.

12. An apparatus comprising the X-ray inspection device (200) according to claim 5 and the processing device (300) according to claim 11, wherein, The X-ray inspection apparatus (200) is configured to provide the second detector data and the first detector data to the processing apparatus (300) based on scanning an inspection object (216), and is connected to the processing apparatus (300) for data communication with it.

13. A computer program product comprising instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 6-10.

14. A computer-readable data carrier comprising the computer program product according to claim 13.

Images