Dual-energy detector and method for improving image data generated therewith

Description

Technical field

[0001] The present disclosure generally relates to non-destructive inspection of objects using dual-energy X-ray inspection to locate target objects. More particularly, the present disclosure relates to a method for enhancing detector data provided by a dual-energy detector. background

[0002] The following introductory description serves only to provide a better understanding of the technical relationships described here and should in no way be understood as acknowledged state of the art unless it is expressly marked as such.

[0003] The discrimination of materials using dual-energy techniques is generally known; see, for example, S. Kolkoori et al., "Dual High-Energy X-ray Digital Radiography for Material Discrimination in Cargo Containers," 11th European Conference on Non-Destructive Testing, 2014, Prague, Proceedings. The dual-energy technique is based on the idea that the energy dependence of a material's attenuation coefficient is a property of the material. Therefore, energy-resolved (in the sense of spectrally selective) measurements can be used to acquire detector data required for a quantitative calculation of material properties. In practice, these energy-resolved measurements can be obtained by taking two X-ray images with different X-ray spectra.The use of a so-called dual-energy detector is established; its special design essentially provides two energy channels, allowing the required high-energy (Hi) and low-energy (Lo) detector data to be obtained with one measurement.

[0004] EP 1 010 021 B1 discloses a dual-energy X-ray detector. The detector comprises a Lo detector array with first Lo detector elements that respond to X-ray radiation in a first low-energy energy range to provide a first radiation sensitivity, and a Hi detector array with second Hi detector elements that respond to radiation in a second high-energy range to provide a second radiation sensitivity. The Lo detector array and the Hi detector array are both arranged one behind the other transversely to the scanning direction of an object such that the first detector elements and the second detector elements can receive the X-ray radiation to be measured independently of one another.

[0005] Figure 1shows a cross-sectional view of an embodiment of a detector 200 known from EP 1 010 021 B1. X-rays are symbolically shown radiating from top to bottom. The detector 200 consists of a circuit board 210, two interfaces 215A and 215B, a Lo ground 220, a Lo photodetector row 230 with Lo photodiodes, a Lo wiring 235, a Lo crystal 240 as a scintillator, a Lo coating 245, a filter 250, a Hi ground 260, a Hi photodetector row 270 with Hi photodiodes, a Hi wiring 275, a Hi crystal 280 as a scintillator and a Hi coating 285. The circuit board 210 thus carries two detector rows 230 and 270 and the interfaces 215A and 215B.

[0006] The gold-plated copper Lo ground 220 connects a common cathode of the photodiodes of the Lo photodetector array 230 as a common ground. The Lo photodetector array 230 responds to Lo X-rays by using the Lo crystal 240 as a scintillator to convert X-ray photons of the Lo range into photons of visible light that can be detected by the photodiodes. The Lo photodiodes detect the light generated in the respective detector element according to the intensity profile of the Lo crystal 240 as a scintillator and generate a corresponding analog signal. The analog signal from each individual photodiode is routed to the data processing board (not shown) via interface 215A. The Lo wiring 235 connects the outputs of the individual photodiodes to the corresponding terminals of the Lo interface 215A.

[0007] The filter 250, made of a material with a high atomic number, such as silver, gold, or copper, enhances the detector array's ability to detect Hi radiation. The gold-plated copper Hi ground 260 connects the cathodes of the photodiodes of the Hi photodetector array 270 as a common ground. Each of the photodiodes responds to Hi X-rays and, similarly to the Lo photodetector array, converts light generated according to the intensity profile of the scintillator material, corresponding to the detected Hi radiation, into an analog signal. The analog signal from each photodiode is then passed to the data processing board via interface 215B.

[0008] The number of photodiodes, the spacing between them, and the dimensions of the photodiodes in each of the photodetector rows define the scanning resolution in the corresponding scan line; these parameters are the same for the Hi photodiode row 270 and the Lo photodiode row 230.

[0009] In practice, the well-known dual-energy detector can be used to acquire the two measured detector data points required for material discrimination per pixel for an inspection object, line by line, in parallel. However, when observing a specific object line, the high-level detector data and the low-level detector data are not acquired simultaneously, but sequentially, i.e., at different times. The specific object line is first scanned by the low-level detector line. Then, after the object has been moved one detector line width further, the specific object line is scanned by the high-level detector line (or vice versa, depending on the arrangement of the two detector lines in the object's transport direction).

[0010] DE 10 2011 053 971 A1, US 9 885 801 B2, US 2014 01 98 899 A1 and US 5 841 833 A each show detector rows with detector elements of different spectral sensitivity arranged alternately therein.

[0011] EP 2 960 686 A1 and US 2010 / 172464 A1 also each show a dual-energy X-ray detector with detector rows in which detector elements of different spectral sensitivity are arranged alternately, for use in a dual-energy X-ray method in a CT X-ray system which operates according to the computed tomography method.

[0012] US 5,841,832 A discloses a dual-energy X-ray detector for use in a dual-energy X-ray process, comprising a detector array arranged transversely to the scanning direction, consisting of low-energy and high-energy detector elements that respond differently to an X-ray spectrum, each arranged alternately in each of the detector arrays and adjacent detector arrays. The low-energy and high-energy detector elements detect X-rays independently of one another and generate corresponding low-energy and high-energy detector data.Furthermore, a method for correcting these low-energy and high-energy detector data is shown, wherein missing low-energy detector data for the location of a high-energy detector element are generated by calculating an interpolation value of measured low-energy detector data from low-energy detector elements adjacent to this high-energy detector element from a detector row or from a detector row and the adjacent detector row. Summary

[0013] In a detector row in which Hi-detector elements and Lo-detector elements are arranged alternately next to one another in one and the same row, an inspection object can be scanned with a desired resolution using the detector elements arranged next to one another transversely to the scanning direction. For each detector location, i.e. pixel, only either Hi-detector data or Lo-detector data are measured and the detector data not recorded at the location of a detector element can be calculated by interpolation according to US 5 841 832 A, so that the dual-energy information required for material discrimination is available.

[0014] It is an object of the invention to propose an improved method for correcting the first and second detector data acquired in this way with such a dual-energy (X-ray) detector.

[0015] The object is achieved with the features of the method of independent claim 1. Further embodiments are specified in claims 2 to 9.

[0016] For a better understanding, the dual-energy (X-ray) detector and the X-ray inspection system are first explained, by means of which the detector data to be corrected with the method are generated.

[0017] The dual-energy detector as a first aspect comprises: a detector line arranged transversely to a scanning direction formed from first

[0018] Detector elements and second detector elements, each arranged alternately next to one another in the detector row. The first and second detector elements detect X-rays independently of one another and each generate only first detector data or second detector data. This means that, based on the respective preset sensitivity for X-rays of the X-ray spectrum, the first and second detector elements each selectively detect only corresponding first detector data of the X-ray spectrum or second detector data of the X-ray spectrum.

[0019] For example, the first detector elements can be Lo detector elements, and the second detector elements can be Hi detector elements. This means that the first detector elements respond to X-rays in a Lo range of the X-ray spectrum, and the second detector elements respond to X-rays in a Hi range of the X-ray spectrum. The second detector elements can also be designed to respond to X-rays in a larger range of the entire X-ray spectrum, but at least in the Hi range of the X-ray spectrum.

[0020] The first detector data of the X-ray spectrum of the first detector elements primarily provide material information regarding how well a irradiated material allows low-energy X-rays to pass through. The second detector data of the X-ray spectrum of the second detector elements primarily provide material information regarding how well a irradiated material allows high-energy X-rays to pass through.

[0021] When an inspection object is irradiated with X-rays containing at least X-rays in the Lo range and in the Hi range of the detector elements, the detector arrangement can be used to record the X-ray intensities associated with each detector element as detector data.

[0022] The detector data provided by the dual-energy detector can basically be an analog or digital detector signal.

[0023] The X-ray spectrum is essentially determined by the X-ray source (X-ray generator) used in a conventional manner. That is, the term X-ray spectrum refers here to the spectrum of X-rays emitted by the X-ray source.

[0024] An inspection object can be scanned line by line using the scanning direction transverse to the scanning direction, which usually corresponds to the transport direction of the inspection object through an inspection system with respect to the detector row arranged in the inspection system, and thus a plurality of first and second detector data can be acquired line by line. As a result, for each pixel of a fluoroscopy image derived from the acquired detector data, either only first or only second detector data is acquired. This means that for each pixel, initially only one type of the two types of detector data required according to dual-energy technology—namely, low detector data and high detector data—is available.

[0025] The first detector elements and second detector elements may, for example, each contain photodiodes with a scintillator material that is selected and dimensioned to adjust the predetermined sensitivity for X-rays of the X-ray spectrum.

[0026] The first detector elements and the second detector elements can be arranged on the same side of a substrate, e.g. in the form of a circuit board, of the detector arrangement or on different sides of the substrate of the detector arrangement.

[0027] As mentioned above, the measured detector data of each individual detector element initially corresponds to an image pixel of an X-ray image derived from the detector data. Therefore, the missing, i.e., unmeasured detector data, are calculated for each pixel using interpolation.

[0028] For image display based on the detector data, each pixel is colored according to the result of the material discrimination based on the well-known dual-energy technique.

[0029] The brightness (luminance) of each pixel can be derived from the Hi-detector data. This can result in a colored X-ray image, for example, a 3-color representation, for color-displaying the objects in the X-ray image according to material classes. With color representation according to material classes, the objects contained in the inspection object are colored depending on the material they are predominantly made of, so that a viewer can immediately recognize an optical assignment to predetermined material classes in the colored X-ray image. For example, metallic objects or organic materials can be highlighted using different colors and more easily distinguished from one another.

[0030] The X-ray inspection system as a second aspect is at least configured for transporting inspection objects in a scanning direction through the X-ray inspection system to carry out an imaging inspection method for non-destructive inspection of the inspection objects and to provide detector data describing the inspection objects based on detecting X-rays penetrating the inspection objects by means of the dual-energy detector according to the first aspect.

[0031] The inventor has found that the following image defects, among others, occur in colored fluoroscopy images generated based on the detector data acquired with the dual-energy detector according to the first aspect in the X-ray inspection system according to the second aspect. (a) False colors (false color information) occur particularly at the edges of objects in the coloring according to material classes derived from the detector data. (b) For small structures where the absorption of the X-rays alternates between high and low values, false colors occur in the coloring according to material classes derived from the detector data. This results in a structure shown in the image that does not correspond to the structure actually present in the object being inspected. An example of such an incorrectly represented structure is integrated components with a BGA package, in which spherical connections are arranged in a grid on one mounting side for SMD attachment to a circuit board (BGA is derived from the English term Ball Grid Array).(c) In areas of an inspection object where a homogeneous material is present, incorrect alternating light-dark structures (luminance fluctuations) occur in the coloring according to material classes derived from the detector data. (d) Incorrect colors are observed in the coloring according to material classes derived from the detector data, which the inventor suspects are caused by different absorption properties of edge pixels of the dual-energy X-ray detector.

[0032] With the method according to the invention, some of the above-mentioned errors in the color representation according to material classes derived from the detector data acquired by means of the dual-energy detector according to the first aspect can be at least reduced or eliminated.

[0033] It should be noted that features and details defined in connection with the detector data correction method naturally apply correspondingly to a system consisting of the X-ray inspection system of the second aspect with the dual-energy (X-ray) detector of the first aspect and a processing device explained below as the third aspect for implementing the method, and vice versa. Therefore, to avoid repetition, reference is made to the individual aspects.

[0034] Finding the technical solution proposed here required several technical insights and considerations. In particular, it was recognized that the observed image errors are related to the specific features of the systematic juxtaposition of Hi-detector elements and Lo-detector elements in the detector row, explained below. The measures proposed here are therefore essentially based on the following insights and technical considerations of the inventor for the X-ray inspection system of the second aspect with a detector of the first aspect. (a) The inventor has recognized that the X-rays detected by two adjacent detector elements at different solid angles lead to false color information in the derived X-ray image, particularly at edges in the inspection object. (b) The inventor has recognized that for small structures in the inspection object, where the absorption of the X-rays alternates between high and low values, erroneous color information occurs because either only one Lo detector element or only one Hi detector element detects the small structure, and thus, due to interpolation, an incorrect structure is concluded overall. For example, integrated components with the aforementioned BGA package have proven problematic because it can happen that either the Lo detector elements or the Hi detector elements always detect a spherical connection and the other detector elements do not, which ultimately leads to the aforementioned image error.(c) The inventor has recognized that, in the case of homogeneous materials in an inspection object, the different absorption properties of the adjacently arranged Lo and Hi detector elements can lead to an alternating light-dark structure (also brightness or luminance fluctuations) in the X-ray image derived from the detector signal. (d) The inventor has recognized that if the Lo and Hi detector elements are arranged in photodiode rows (photo diode arrays, PDAs), direct crosstalk can occur between the different channels. Since at the edge of the photodiode row (i.e. at the edge of the detector row) either a Lo or Hi detector element has only one immediately adjacent detector element, the absorption properties of the detector elements located at the two edges of the detector row are different.This also appears to lead to incorrect coloring of corresponding pixels in the X-ray image derived from the detector signal.

[0035] Therefore, a solution was sought to reduce the image disturbances discussed above or to correct them in such a way that the X-ray image displayed to an operator, colored according to material classes, can be easily evaluated.

[0036] The invention thus relates to a method for correcting detector data of the X-ray inspection system according to the second aspect with the dual-energy X-ray detector according to the first aspect.

[0037] The following image enhancement measures are preceded: The color of a pixel in the colored fluoroscopy image derived from the detector data is derived based on material discrimination by determining the Z-eff from the high-energy (Hi) detector data and low-energy (Lo) detector data for a pixel.

[0038] The brightness (luminance) of a pixel is determined based on one type of detector data, preferably the Hi detector data, but the Lo detector data can also be used.

[0039] The following measures can be used individually to correct a specific problem, or all together. In principle, all of the measures suggested here can be applied simultaneously or individually. Each of the image-enhancing measures has been recognized as self-adjusting, as each measure "virtually" only has a significant effect at its intended location in the image and leads to no significant changes (disturbances) elsewhere.

[0040] It is proposed here to correct the detector data provided by the inspection system according to the second aspect with the dual-energy (X-ray) detector of the first aspect using a method, the method comprising the following steps: In a first step (a1) of the method, missing second detector data for the location of a first detector element are first derived. For this purpose, an interpolation value is first determined from measured second detector data of second detector elements adjacent to this first detector element. Then, a correction value is determined based on first detector data measured with this first detector element and adjacent measured first detector data. Finally, the initially determined interpolation value is corrected with this correction value.

[0041] Determining the correction value in step (a1) may comprise: forming the difference between n times the detector data measured with the first detector element and the sum of n measured first detector data adjacent in the detector row, where n is an integer, even number with n>=2

[0042] In addition, a second step (a2) of the method derives missing first detector data for the location of a second detector element. For this purpose, an interpolation value is first determined from measured first detector data of first detector elements adjacent to this second detector element. Then, a correction value is determined based on second detector data measured with this second detector element and adjacent measured second detector data. Finally, the initially determined interpolation value is corrected with the correction value.

[0043] Determining the correction value in step (a2) may comprise: forming the difference between n times the detector data measured with the second detector element and the sum of n measured second detector data adjacent in the detector row, where n is an integer, even number with n>=2.

[0044] The above measure is a modified interpolation for determining the detector data missing at a particular detector location, which ultimately leads to improved material discrimination quality. This measure avoids image distortions caused by false colors at edges in the inspection object and false colors in areas with fine structures of an inspection object, which can occur due to the special structure of the dual-energy detector of the first aspect. In particular, it has been found that this measure does not lead to any significant changes in a homogeneous material. In areas with high-frequency structural changes, false colors due to incorrect scanning with the first or second detector elements are avoided or at least reduced.

[0045] The correction value can, for example, be a correction factor. Correcting the interpolation value with the correction value can then be performed by multiplying the interpolation value by the correction factor.

[0046] Determining the correction factor in step (a1) can comprise: forming the difference between 1 and a quotient of a difference between the two second detector data measured adjacent to the left and right sides in the detector row and the sum of the two second detector data measured adjacent to the left and right sides in the detector row. Accordingly, determining the correction factor in step (a2) can comprise: forming the difference between 1 and a quotient of a difference between the two first detector data measured adjacent to the left and right sides in the detector row and the sum of the two first detector data measured adjacent to the left and right sides in the detector row.

[0047] A further measure of the invention concerns the correction of brightness or light-dark fluctuations. To determine the brightness (luminance) of the pixels of the final fluoroscopy image, normally only one type of detector data is used uniformly for all pixels, i.e., either only the high-level detector data or only the low-level detector data. Since, in the simplified detector row of the first aspect, the detector data required to determine the brightness of the pixel is missing as measured data for every second pixel, the brightness information for all pixels is initially based half on interpolated detector data (i.e., detector data determined by calculation compared to measured detector data).

[0048] The following additional measure aims to improve the brightness value assigned to a pixel based on (calculated) detector data. The following steps are proposed: (b1) Determination of material values ​​for all pixels based on the detector data actually measured for each pixel and the detector data determined (by calculation) for each pixel. The material values ​​assigned to the pixels define a material information matrix by storing a Z-effective value as a material value in the material information matrix for each pixel corresponding to a detector element. (b2) Filtering of the material values ​​determined for all pixels in the material information matrix using a mean-based or median-based filter operation, which captures NxN material values ​​around each pixel, where N is an integer greater than 3, preferably N=5. "Mean-based" or "median-based" here means that the result can be additionally weighted with a factor or adjusted with an offset.(b3) Determining, at each pixel at which first detector data were measured, the material value determined for that pixel from theoretical second detector data associated with that pixel. "Theoretical detector data" here means that, with the dual-energy technique, if the Z-eff of a radiated material is known and one type of detector data, namely the Hi or Lo detector data, is known, the unknown detector data can be theoretically determined. (b4) Determining a respective brightness value for all pixels based on the measured second detector data or the theoretical second detector data available for the respective pixel.

[0049] The second detector data is preferably Hi-detector data. This measure can correct luminance disturbances, particularly luminance fluctuations. This means that, using the previous measures (discussed above), the material discrimination on which the coloring of the final fluoroscopy image is based can be improved. The inventor has recognized that the material values ​​thus determined, as described above, can be used to determine theoretical Hi-detector data, which then determine the luminance of the respective pixel.

[0050] As already mentioned, this luminance correction measure can also be used based on low-level detector data as second detector data. Since, with a known material value for a pixel, the interpolated detector data for this location and the measured detector data for this location are related by a factor, low-level detector data can also be used for the luminance correction, which is preferably based on high-level detector data.

[0051] The above measure can be used to avoid light-dark fluctuations in image areas where the inspection object actually has a homogeneous material and should therefore be uniformly bright in the fluoroscopy image.

[0052] A fourth aspect relates to a processing device for correcting detector data, wherein the processing device comprises a computer unit configured to carry out a method according to the third aspect.

[0053] A fifth aspect relates to a system comprising the X-ray inspection system of the second aspect, which is configured to forward the acquired detector data to a processing device according to the fourth aspect. For this purpose, the X-ray inspection system can be connected to the processing device for data communication, for example, wired or wireless.

[0054] A sixth aspect relates to a computer program product comprising a computer program having software means for carrying out a method according to the third aspect when the computer program is executed on a computer, in particular on a processing device according to the fourth aspect.

[0055] A seventh aspect relates to a data carrier containing a computer program product according to the third aspect.

[0056] An eighth aspect relates to a data stream comprising electronically readable control signals which can interact with a programmable computer, in particular a processing device according to the fourth aspect, such that when the computer executes the electronically readable control signals, the computer performs a method according to the third aspect. Short Description of the drawing figures

[0057] Further advantages, features, and details of the solution(s) proposed here emerge from the following description, in which exemplary embodiments are described in detail with reference to the drawings. The features mentioned in the claims and in the description can be significant individually or in any combination. Likewise, the features mentioned above and those further explained here can be used individually or in combinations. Parts or components with similar functions or that are identical are sometimes provided with the same reference numerals. The terms "left," "right," "top," and "bottom" used in the description of the exemplary embodiments refer to the drawings in an orientation with normally legible figure designations or normally legible reference numerals.The embodiments shown and described are not intended to be exhaustive, but rather are exemplary in nature to explain the solution proposed here. The detailed description serves to inform those skilled in the art; therefore, known structures and methods are not shown or explained in detail in order not to obscure the description. Figure 1 shows a cross-sectional view of a prior art detector with a row of first detector elements and a row of second detector elements, wherein the rows are arranged one behind the other in the scanning direction. Figure 2 shows a cross-section through a novel detector row, wherein Lo detector elements and Hi detector elements are arranged alternately next to each other in the detector row. Figure 3A illustrates the matrix structure of acquired detector data, which is generated by means of the detector row of the Figure 2by scanning an inspection object through X-rays line by line. Figure 3B illustrates that from the acquired detector data of the matrix structure of the Figure 3a , a matrix containing acquired and calculated detector data for each pixel is calculated. Figure 4 shows an example of an X-ray inspection system with a detector with a detector row of Figure 2 in a side cross-sectional view. Figure 5 illustrates the determination of missing detector data for the detector row of the Figure 2 for an object with a homogeneous material distribution. Figure 6 illustrates interpolation errors in the determination of missing detector data for the detector row of the Figure 2 for an object with small-scale fluctuations in the material distribution. Figure 7 illustrates interpolation errors in the determination of missing detector data for the detector row of the Figure 2at the edge of an object, which represents a jump in the material distribution. Figure 8 illustrates a luminance correction measure that preserves the smallest structures in the inspection object. Detailed description of implementation examples

[0058] Figure 1 shows the cross-sectional view of the dual-energy detector 200 of the prior art discussed in the introduction, which is known from EP 1 010 021 B1.

[0059] Figure 2 shows, in a highly simplified representation, a cross-section through a detector row 130 of a dual-energy X-ray detector 100 proposed here according to an exemplary embodiment. In the detector row 130, low-energy (Lo) detector elements 132 and high-energy (Hi) detector elements 134 are arranged alternately next to one another.

[0060] The detector row 130 essentially consists of a photodiode array (PDA) with photodiodes 137 arranged on a carrier board 120 as a substrate. The photodiodes 137 are each alternately provided with an associated scintillator 133 or 135 in the detector row 130.

[0061] By using special materials in the scintillators 133, 135 and / or dimensioning the layer thickness (in the y-direction) of the respective scintillators 133, 135, the response behavior of the respective photodiode 137 implements a Lo detector element 132 or a Hi detector element 134.

[0062] For example, in one implementation, the dimensions of a detector element can be 0.8 mm (in the z-direction) by 0.8 mm (in the row direction, i.e., the x-direction). A typical detector row can have a total length of 80 mm in the row direction (x-direction), so the row consists of approximately 1,000 detector elements. In principle, the dimensions of the detector elements and the length of the detector row can be freely selected.

[0063] For example, as in the Figure 2 Simplified (and therefore exaggerated, ie not to scale), the Lo detector elements 132 have a scintillator 133 made of a special scintillator material L with a first layer thickness (which is determined in the direction of incidence for the X-rays, ie in the y-direction). The Hi detector elements 134 have a scintillator 135 made of another special scintillator material H with a second layer thickness. As shown in the Figure 2shown, the first layer thickness is smaller than the second layer thickness.

[0064] It should be noted that the scintillator materials L and H can, in principle, also be identical, in which case the response behavior is determined exclusively by the layer thickness. It is also possible for the layer thicknesses of the Hi and Lo detector elements 132, 134 to be identical, in which case the response behavior is determined exclusively by the different scintillator materials L and H.

[0065] For example, YAG (yttrium aluminum garnet), YGAG (yttrium gadolinium aluminum garnet), GOS (gadolinium oxysulfide) or materials that have a physically similar effect to X-rays can be used as scintillator materials.

[0066] For the Lo scintillator, first layer thicknesses of the order of magnitude less than 1 mm (e.g. 0.45 mm) are preferred, and for the Hi scintillator, second layer thicknesses of the order of magnitude greater than 1 mm (e.g. 1.4 mm) are preferred.

[0067] At the Figure 2 In the detector row shown, all detector elements 132, 134 are arranged on the same side, namely the top side 121 of the carrier board 120. It should be noted that (in a variant not shown) the Lo detector elements 132 can be arranged on the top side 121 of the carrier board 120, while the Hi detector elements 134 are located on the bottom side 122 of the carrier board 120, or vice versa.

[0068] For orientation and reference, the Figure 2a coordinate system is plotted, according to which the longitudinal direction of the detector line 130 runs in the x-direction, the direction of X-rays RX incident on the detector elements 132, 134 to be detected (simply shown as a bundle of arrows) runs in the y-direction and the scanning direction of the detector line runs in the z-direction. The scanning direction corresponds to the transport direction TR of inspection objects through the Figure 4 X-ray inspection system 1 shown as an example. What is important in this context is that the longitudinal direction (x-direction) of the detector row 130 proposed here is arranged transversely to the scanning direction (z-direction).

[0069] Figure 3A illustrates a matrix structure of acquired detector data obtained by means of the detector row 130 of the Figure 2 by scanning an inspection object line by line through X-rays.

[0070] In the Figure 3AFor orientation and reference, the coordinate system of the Figure 2 Accordingly, the detector row 130 generates the detector data matrix 300 when scanning row by row.

[0071] The detector data matrix 300 is composed of rows 330a, 330b, 330c, ..., 330n in the z-direction. Due to the alternating arrangement of the Lo and Hi detector elements 132, 134, the detector data matrix 300 displays columns 340a, 340b, 340c, ..., 340m, each consisting only of measured Lo detector data or Hi detector data. This means, as already explained, the simplification in the structure of the detector row 130 with respect to one type of detector element each was "acquired" by undersampling. This undersampling can, however, be compensated for by reconstruction, for example, by interpolation, of the type of detector data missing at a particular pixel.

[0072] Figure 3B illustrates that from the acquired detector data matrix the Figure 3A , a complete detector data matrix is ​​derived, containing the acquired (measured) and determined (calculated) detector data for each pixel. For example, the missing type of detector data for a pixel can be calculated and thus determined by interpolating the two neighboring measured detector data.

[0073] Figure 4 shows an example of an X-ray inspection system 1 with a detector 100 with a detector row 130 of the Figure 2in a lateral cross-sectional view. The X-ray inspection system 1 has radiation protection curtains 3a, 3b, one of which is arranged at each entrance and exit of a radiation tunnel 2 of the X-ray inspection system 1. Between the radiation protection curtains 3a, 3b, within the radiation tunnel 2, there is a radiation area 4 in which at least one radiation source 5 (e.g., an X-ray tube with a collimator) and the detector 100 aligned therewith are arranged. A transport device 8, for example, a sliding belt conveyor, is used to transport a piece of luggage 7 as an inspection object in the transport direction TR through the radiation tunnel 2. The detector 100, and thus also the detector row 130, is L-shaped and arranged with its longitudinal direction transverse to the transport direction TR, so that the transport direction TR corresponds to the scanning direction of the inspection objects 7.

[0074] Figure 4 is also a highly simplified block diagram of a system 400, which essentially consists of the X-ray inspection system 1 and a processing device 410, which is set up to carry out the methods proposed here for correcting detector data which are obtained with an X-ray inspection system, for example the X-ray inspection system 1 of Figure 4 which are equipped with a dual-energy detector, for example the dual-energy detector 100 of the Figure 2 , is equipped.

[0075] The processing device 410 can, as in the Figure 4illustrated, be part of a control device 420 of the X-ray inspection system 1. The corrected or compensated detector data can be used in the control device 420 or a separate image processing device connected thereto to generate the desired fluoroscopic image colored based on material classes, which is displayed to an operator on a screen (not shown) in a manner known per se. The processing device 410 can, in principle, also be located completely separate from the X-ray inspection system, for example, at a central location where the raw detector data from several inspection systems converge and are processed centrally there. However, this makes no difference to the effectiveness of the corrective measures proposed here.

[0076] For example, the processing unit 410 can also be part of the dual-energy detector unit 100, so that the data generated by the detector unit 100 is immediately enhanced at the detector according to the measures proposed here. This could make the dual-energy detector unit 100 proposed here fundamentally compatible with existing X-ray inspection systems with conventional, more complex detector units. As a result, in otherwise identical X-ray inspection systems, the dual-energy detector unit 100 proposed here, in conjunction with the enhancement measures for the detector data, can reduce system costs while maintaining a similarly high image quality.

[0077] The discussion of the reconstruction measures of detector data missing at the location of a detector element illustrated in Figures 5-9 is preceded by an explanation of the symbols used.

[0078] In each of Figures 5-9, the detector row 130 is shown as described in connection with the Figure 2 explained. For simplified representation, the alternating Lo detector elements are labeled L and the Hi detector elements are labeled H. In each figure, the X-rays RX incident from an X-ray source are symbolically indicated by a bundle of arrows, with the X-rays RX running from bottom to top in the figures, each passing through a test object or free space and then striking the detector row 130, from whose detector elements 132, 134 Lo detector data or Hi detector data are acquired for one pixel each.

[0079] The detector data are represented by respective symbols. Hi-detector data are triangles and Lo-detector data are circles. The symbol for the detector data acquired by a detector element, i.e., the detector data actually measured with the respective detector, is a solid symbol. A type of detector data reconstructed for a specific detector is represented by the corresponding symbol with a dashed border. If detector data calculated by interpolation do not correctly match the actual circumstances, the corresponding corrected detector data is represented by the corresponding symbol, the area of ​​which is hatched. In addition, an arrow indicates the direction in which the inappropriate interpolation value should be corrected.

[0080] Figure 5 first illustrates the determination of missing detector data for detector row 130 of the Figure 2 for an object G with a homogeneous material distribution.

[0081] According to the boundary conditions, all actually measured Lo detector data (triangles) are the same, as are all actually measured Hi detector data (circles). Consequently, an interpolation of the actually measured detector data adjacent to the defect leads to an interpolation value that corresponds to the physical boundary conditions. The interpolation values ​​calculated in this way for the detector data matrix 300 ( Figure 3B ) are suitable for material discrimination according to dual-energy technology and do not lead to incorrect coloring of pixels.

[0082] In the Figure 5 Two examples of reconstruction of missing detector data are shown. In the left part of the Figure 5the detector element 132 as a Lo detector element has acquired the Lo detector data 500. The unmeasured Hi detector data are calculated by interpolating the immediately adjacent measured Hi detector data 501 and 503. This leads to the equally large interpolation value 505, since the Hi detector data 501 and 503 are of the same size. In the right area of ​​the Figure 5 the detector element 134 as a Hi detector element has acquired the Hi detector data 510. The unmeasured Lo detector data are calculated by interpolating the immediately adjacent measured Lo detector data 511 and 513. This leads to the equally large interpolation value 515, since the Lo detector data 511 and 513 are of equal size. As already mentioned above, the situation of Figure 5 to no false colors in the derived fluoroscopy image colored according to material classes.

[0083] Figure 6illustrates interpolation errors in determining missing detector data for detector row 130 of the Figure 2 for an object G with small-scale fluctuations in the material distribution.

[0084] The physical conditions in the Figure 6 differ essentially in that the small-scale fluctuations in the material properties of the inspection object mean that the interpolation of neighboring actually measured detector data cannot correctly reconstruct the missing detector data.

[0085] In the Figure 6 An example of such an erroneous reconstruction of missing detector data is shown. In the left part of the Figure 6The detector element 132, as the Lo detector element, has acquired the Lo detector data 600. The unmeasured Hi detector data are to be calculated again by interpolating the immediately adjacent measured Hi detector data 601 and 603. Since the Hi detector data 601 and 603 are different, this leads to an average value of the two Hi detector data 601 and 603 adjacent to the missing Hi detector data due to the interpolation. The interpolation value 605 obtained in this way is significantly too high. The target value 607 for correct reconstruction is also plotted and, due to the framework conditions, corresponds to the detector data 603. The arrow COR indicates the direction in which a correction or compensation measure would have to correct the interpolation value; ideally to the value of the detector data 603.

[0086] The situation of the Figure 6therefore leads to the generation of false colors in the derived fluoroscopy image colored according to material classes.

[0087] As a corrective measure, it is proposed in principle to first calculate missing detector data of a type by interpolation from n detector data of the same type that are closest in the detector row, where n is an even number greater than or equal to 2, i.e. n = 2, 4, 6, .... This interpolation value is then corrected with a suitable correction value.

[0088] The following approach has proven effective for determining the correction value. The detector data of the other type (i.e., the complementary energy range) actually measured at the interpolation point are compared with their ambient values, and based on this, a correction of the interpolation value is performed in the correct direction (COR). The more inhomogeneous and the farther away from the ambient mean the values ​​are, the larger the correction value should be.

[0089] It is good practice to calculate the correction value as n times the detector data (i.e., of the complementary / other type) actually measured at the interpolation point minus the n actually measured complementary detector data closest to the interpolation point in the detector row.

[0090] For this purpose, the Figure 6 An example is illustrated in the right-hand part. The Hi detector data 610 were measured at the location of the Hi detector element 134. The complementary Lo detector data missing at this location are calculated according to the strategy proposed above, for example, with n=4, as follows.

[0091] First, an interpolation value 617 is calculated based on the measured Lo detector data 600, 611, 613, 615 closest to n=4 in detector row 130. The correction value is calculated as the difference between 4 times the Hi detector data 610 measured by Hi detector element 134 and the Hi detector data 601, 603, 606, 608 closest to n=4 in detector row 130. Using the correction value thus obtained, the interpolation value 617 is corrected to approximate as closely as possible the correct Lo detector data 620 to be reconstructed.

[0092] The effect of interpolation is to reduce edge discoloration. The fine structures in the illuminated object cause a kind of beating in the color representation, the effect of which can be reduced by mixing in gradient terms or difference terms from neighboring measured detector data.

[0093] Figure 7illustrates interpolation errors in determining missing detector data for detector row 130 of the Figure 2 at the edge of an object, which means a jump in the material distribution.

[0094] As already mentioned in the Figure 6 Interpolation at the edges of objects produces false colors in the derived fluoroscopy image colored according to material classes, since corrupted detector data influences the material discrimination of the dual-energy technology.

[0095] Another correction approach that has proven effective at edges is to first determine an interpolation value at the point of interest (defect) at a location with a detector element of one type (Hi or Lo), based on n nearest measured detector data of the same type in the detector row.

[0096] For example, in the Figure 7at the location of the Hi detector element 134 of the detector parts 130, the interpolation value 705 of the missing Lo detector data is calculated based on the n=2 nearest neighboring actually measured Lo detector data 701, 703; n should generally be an even number greater than or equal to 2. This first calculated interpolation value 705 is incorrect. As can be seen from the physical boundary conditions of the Figure 6 understandable, the interpolation value 705 is too large.

[0097] Therefore, based on empirical tests, it is proposed here to correct the interpolation value 705 with a correction value used as a correction factor. It has been found that a correction factor works well, which is determined as: the difference of 1 and the quotient of the difference between k complementary, actually measured Hi-detector data 709, 711 closest to detector row 130 and the sum of the k complementary, actually measured Hi-detector data 709, 711 closest to detector row 130. This relatively simple calculation rule achieves a very good compensation for interpolation errors at edges and thus avoids the color distortions that occur without correction.

[0098] Figures 8 and 9 illustrates an additional measure for luminance correction to reduce light-dark fluctuations in the fluoroscopy image colored according to material classes.

[0099] The brightness information for each pixel is typically determined based on the Hi-detector data for each pixel. This is particularly useful when the Hi-detector elements are configured to respond to the entire X-ray spectrum of the X-ray source used.

[0100] The following approach has been found to be extremely effective for luminance correction.

[0101] First, the detector data matrix 300* is used to calculate the Figure 3B Material discrimination is performed. For this purpose, the reconstructed detector data may already have been corrected using the above measures. The result is a material data matrix containing material information for each pixel, for example, the determined Z-effective (atomic number).

[0102] This material data matrix is ​​then filtered using an NxN mean or median filter. The result is a filtered material data matrix.

[0103] Using the filtered material data matrix, the theoretical corresponding Hi-detector data can now be calculated in the original detector data matrix 300 for all pixels for which no Hi-detector data is available, based on the measured Lo-detector data and the associated filtered material data, based on the characteristic curve of the detector element, and a corrected brightness value can be determined accordingly. "Characteristic curve of the detector element" describes the absorption properties and material characteristics of the detector element.

[0104] It has been shown that this procedure effectively compensates for brightness fluctuations, but that the smallest structures, such as wires, in the inspection object are preserved in the fluoroscopy image.

[0105] Figure 8illustrates how, using the brightness correction proposed here, fine structures of an object in the fluoroscopy image can be preserved by skillfully correcting the brightness values.

[0106] In the situation of Figure 8 A wire W lies on the homogeneous object G as a microstructure. The wire W lies at the location of the Lo detector element 132, which at this location acquires somewhat more attenuated Lo detector data compared to the surroundings. The Hi detector data missing at this location would lead to inflated Hi detector data 805 due to the homogeneous environment based on the neighboring Hi detector data 801, 803.

[0107] Since the brightness of a pixel in the fluoroscopy image is essentially determined by the Hi-detector data, the wire W would be filtered out in the worst case. With the measure presented above, the Hi-detector data 807 is determined for brightness control based on the actually measured Lo-detector data and the material value (Z-effective) determined for this pixel via the characteristic curve of the detector element. Compared to the other measures proposed to improve material discrimination, this leads to better resolution of the smallest structures in the fluoroscopy image solely through the resulting luminance correction or luminance control (brightness control).

[0108] It should be noted that material discrimination and brightness control are independent of each other and can all be applied together. No particular order is preferred.

[0109] All of the methods presented here work sufficiently well across the entire detector array, essentially up to the edge at a distance equal to the number of detector elements required for correction. However, the tunnel of the X-ray inspection system physically ends at the edges of the array. This makes the edge area of ​​the detector array uninteresting or unimportant for observation.

Claims

1. Method for correcting detector data provided by an inspection system (1), wherein the inspection system (1) comprises a dual-energy X-ray detector (100) for use in a dual-energy X-ray method, having a detector line (130) to be arranged transversely to the scanning direction and consisting of first detector elements (132) and second detector elements (134) that respond differently to an X-ray spectrum, which elements are arranged alternately next to one another in the detector line (130), wherein the first and second detector elements (132, 134) detect X-radiation (XR) independently of each other, and wherein the inspection system (1) is set up for transportation in a scanning direction of inspection objects (7) through the inspection system (1) for carrying out an imaging inspection method for the non-destructive inspection of the inspection objects (7) and for providing detector data describing the inspection objects (7) based on detection of X-rays (XR) passing through the inspection objects (7) with the detector (100), wherein the first detector elements (132) generate first detector data and the second detector elements (134) generate second detector data; wherein the method for correcting the detector data provided by the inspection system (1) is carried out by a processing device (410) for correcting detector data and comprises the following steps: (a1) deriving missing second detector data for the location of a first detector element by: determining an interpolation value of measured second detector data from second detector elements neighboring said first detector element; determining a correction value based on measured first detector data from said first detector element and neighboring measured first detector data; and correcting said interpolation value with said correction value; and (a2) deriving missing first detector data for the location of a second detector element, by: determining an interpolation value of measured first detector data from first detector elements neighboring said second detector element; determining a correction value based on measured second detector data from said second detector element and neighboring measured second detector data; and correcting said interpolation value with said correction value; characterized by a correction of an associated brightness value determined on the basis of specific detector data of a pixel, comprising the steps of: (b1) determining material values for all image points based on the detector data actually measured for each image point and the detector data determined for each image point, the material values associated with the image points defining a material information matrix by storing a Z-effective value as a material value in the material information matrix for each image point corresponding to a detector element; (b2) filtering the material values determined for all image points in the material information matrix by means of a mean-based or median-based filter operation, which in each case detects NxN material values around an image point, where N is an integer greater than 3; (b3) determining, at each pixel at which first detector data has been measured, on the basis of the material value determined for this pixel, the theoretical second detector data associated with this pixel; and (b4) determining a respective brightness value for all pixels based on the measured second detector data or the theoretical second detector data for the respective pixel.

2. The method according to claim 1, wherein in step (b3) the material values determined for all image points in the material information matrix are filtered by means of a mean-based or median-based filter operation, which in each case detects NxN material values around an image point, where N=5.

3. The method according to claim 1 or 2, wherein determining the correction value in step (a1) comprises: forming the difference between the n-fold of the detector data measured with the first detector element and the sum of n neighboring measured first detector data in the detector line; and / or determining the correction value in step (a2) comprises: forming the difference between the n-fold of the detector data measured with the second detector element and the sum of n second detector data measured adjacent in the detector line; wherein n is an even integer with n>=2.

4. The method according to any one of claims 1-3, wherein the correction value is a correction factor and correcting the interpolation value with the correction value comprises multiplying the interpolation value by the correction factor; and wherein determining the correction factor in step (a1) comprises: forming the difference between 1 and a quotient of a difference between the two second detector data measured adjacently on the left and right in the detector line (130) and the sum of the two second detector data measured adjacently on the left and right in the detector line (130); or determining the correction factor in step (a2) comprises: forming the difference between 1 and a quotient of a difference between the two first detector data measured in the detector line (130) on the left-hand side and on the right-hand side and the sum of the two first detector data measured in the detector line (130) on the left-hand side and on the right-hand side.

5. System (400) comprising an inspection system (1) and a processing device (410), the inspection system (400) being set up to provide detector data based on the transillumination of inspection objects (7) to the processing device (410) and being connected to the processing device (410) for data communication; wherein the inspection system (1) comprises a dual-energy X-ray detector (100) for use in a dual-energy X-ray method, comprising: a detector line (130) to be arranged transversely to the scanning direction, consisting of first detector elements (132) and second detector elements (134) that respond differently to an X-ray spectrum, which are arranged alternately next to one another in the detector line (130), the first and second detector elements (132, 134) detecting X-rays (XR) independently of one another; wherein the inspection system (1) is arranged for transportation in a scanning direction of inspection objects (7) through the inspection system (1) for performing an imaging inspection method for non-destructive inspection of the inspection objects (7) and for providing detector data describing the inspection objects (7) based on detection of X-rays (XR) passing through the inspection objects (7) with the detector (100), wherein the first detector elements (132) generate first detector data and the second detector elements (134) generate second detector data; and wherein the processing device (410) is configured to correct the detector data to perform a method according to one of claims 1-4.

6. The system according to claim 5, wherein said first detector elements (132) and second detector elements (134) each comprise a photodiode (137) each provided with a scintillator material (133, 135) selected for adjusting the sensitivity for X-rays of the X-ray spectrum and dimensioned in the thickness to be traversed by the X-radiation.

7. The system according to claim 5 or 6, wherein the first detector elements (132) and the second detector elements (134) are on the same side (121) of a substrate (120), or the first detector elements (132) are located on one side (121) of the substrate (120) and the second detector elements (134) are located on the other side (122) of the substrate (120).

8. Computer program product comprising a computer program having software means for carrying out the method according to any one of claims 1-4 when the computer program is executed on a computer.

9. Transferring a computer program product on a data carrier or a data stream with electronically readable control signals, wherein the computer program product is one according to claim 8.