Calibration method for dual energy spectral images

Abstract

The application provides a kind of dual-energy spectrum image calibration method, the method comprises: obtaining the multiple strip data of each detection unit in the detector of security inspection machine for the article to be inspected, each strip data includes the position of corresponding detection unit and the initial high-low energy data collected by detection unit;For each strip data, find the target area to which the position in the strip data belongs from the pre-set calibration coefficient table, and find the target high-low energy data distribution curve corresponding to the initial high-low energy data in the strip data from the multiple candidate high-low energy data distribution curves in the target area, and determine the calibration coefficient corresponding to the target high-low energy data distribution curve as the target calibration coefficient;Based on the target calibration coefficient, the initial high-low energy data is calibrated to obtain the calibrated high-low energy data;Based on each calibrated high-low energy data, the calibration image of the article to be inspected is determined.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a calibration method for dual-energy spectral images. Background Technology

[0002] In the rail transit sector, dual-energy X-ray security inspection machines have been widely used in airports, train stations, customs, and other locations for detecting prohibited items. These devices collect high- and low-energy data from the items to be inspected and map this data into a pseudo-color image. The pseudo-color image is then used to determine whether the item is organic, inorganic, or a mixture. Furthermore, by determining the equivalent atomic number of the item, a more precise classification of the item can be achieved.

[0003] Due to differences in the structure and hardware performance of the X-ray source (X-ray emission module) and detector (data acquisition module) of security inspection machines, the high- and low-energy data collected when the same item to be inspected is placed in different positions will vary significantly, reducing the accuracy of item classification. In existing technologies, to improve the accuracy of item classification, security images can be calibrated using the machine's background image data, full-scale image data, and original image. This calibration reduces the differences in background data between different detectors, enabling the differentiation of organic, inorganic, and mixed substances.

[0004] However, the above calibration methods still cannot solve the problem of high and low energy data offset of the same item in different spatial locations, resulting in color bands related to pixel positions still existing in the calibrated image, which makes the image quality of the calibrated image low. Summary of the Invention

[0005] This invention provides a calibration method for dual-energy spectral images to address the shortcomings of existing technologies where the high and low energy data of the same object at different spatial locations are offset, resulting in color bands related to pixel positions still appearing in the calibrated image, thus leading to low image quality. This invention aims to improve the image quality of the calibrated image.

[0006] This invention provides a method for calibrating dual-energy spectral images, comprising:

[0007] The system acquires multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit.

[0008] For each strip data, the target region to which the position in the strip data belongs is found from a preset calibration coefficient table, and the target high-low energy data distribution curve corresponding to the initial high-low energy data in the strip data is found from multiple candidate high-low energy data distribution curves in the target region. The calibration coefficient corresponding to the target high-low energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high-low energy data distribution curves of calibrators of different materials in each region, and calibration coefficients corresponding to the high-low energy data distribution curves in each region. The calibration coefficients are used to map the high-low energy data on the corresponding high-low energy data distribution curves to the reference region.

[0009] Based on the target calibration coefficient, the initial high and low energy data are calibrated to obtain calibrated high and low energy data;

[0010] Based on the calibrated high and low energy data, the calibration image of the item to be inspected is determined.

[0011] According to a dual-energy spectral image calibration method provided by the present invention, the step of calibrating the initial high- and low-energy data based on the target calibration coefficient to obtain calibrated high- and low-energy data includes:

[0012] Based on the following formulas (1) and (2), the initial high and low energy data are calibrated to obtain the calibrated high and low energy data:

[0013] (1)

[0014] (2)

[0015] in, This indicates the calibrated high-energy data. This represents the initial high-energy data. This indicates the calibrated low-energy data. The initial low-energy data is represented by k1, c1, r1, and d1, which represent the target calibration coefficients.

[0016] According to a dual-energy spectral image calibration method provided by the present invention, the method further includes:

[0017] For calibration objects of different thicknesses and materials placed on the conveyor belt of the security inspection machine, the calibration objects of different thicknesses are divided into multiple areas along a direction perpendicular to the running direction of the conveyor belt;

[0018] For each region, high and low energy calibration data of calibration objects of different thicknesses collected by the detection units within the region are obtained;

[0019] Based on the high and low energy data of the calibration material with different thicknesses, the distribution curve of the high and low energy data of the calibration material in the region is determined;

[0020] The high and low energy data distribution curves of the calibration materials of different materials in each region are stored in the calibration coefficient table.

[0021] According to a dual-energy spectral image calibration method provided by the present invention, the step of determining the high and low energy data distribution curve of the calibration material in the region based on the calibration high and low energy data of calibration materials of different thicknesses includes:

[0022] For each material of the calibration object, a polynomial fitting algorithm is used to perform polynomial fitting on the calibration high and low energy data of the calibration object at different thicknesses in the region, so as to obtain the high and low energy data distribution curve of the calibration object in the region.

[0023] According to a dual-energy spectral image calibration method provided by the present invention, the step of acquiring calibration high and low energy data of calibration materials of different thicknesses collected by the detection unit in the region includes:

[0024] Obtain the original calibration high and low energy data of calibration objects of different thicknesses collected by the detection unit within the region;

[0025] Full-scale calibration is performed on each of the original calibration high and low energy data to obtain the calibration high and low energy data.

[0026] According to a dual-energy spectral image calibration method provided by the present invention, the calibration object is placed along the width direction of the conveyor belt, and the length of the calibration object is greater than or equal to the width of the conveyor belt.

[0027] According to the present invention, a calibration method for a dual-energy spectral image is provided, wherein the calibrator includes at least two of the following: iron, copper, aluminum, plexiglass, PVC, water in a plastic container, and alcohol in a plastic container.

[0028] According to a dual-energy spectral image calibration method provided by the present invention, the method further includes:

[0029] The reference region is determined from the plurality of regions;

[0030] For any non-reference region, the high and low energy data distribution curve of the calibration object of any material in the non-reference region is approximated by the high and low energy data distribution curve of the calibration object of the same material in the reference region through linear transformation, so as to obtain the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of the material in the non-reference region.

[0031] The calibration coefficients corresponding to the high and low energy data distribution curves of different materials in each of the non-reference regions are stored in the calibration coefficient table.

[0032] According to a dual-energy spectral image calibration method provided by the present invention, the method involves approximating the high- and low-energy data distribution curves of a calibration object of any material in the non-reference region to the high- and low-energy data distribution curves of a calibration object of the same material in the reference region through linear transformation, thereby obtaining calibration coefficients corresponding to the high- and low-energy data distribution curves of the calibration object of the material in the non-reference region. The method includes:

[0033] According to formula (3), determine the calibration coefficient corresponding to the high and low energy data distribution curve of the material in the non-reference area:

[0034] (3)

[0035] in, , representing the polynomial of the high-energy data distribution curve of the calibration material of the same material within the reference region. This represents the high-energy data of the calibrators within the reference area. The polynomial coefficients represent the high-energy data distribution curves of calibrators of the same material within the reference region. , The polynomial coefficients represent the high and low energy data distribution curves of the material in the non-reference region, where k, c, r, and d represent calibration coefficients.

[0036] According to a dual-energy spectral image calibration method provided by the present invention, the method further includes:

[0037] Based on the calibration coefficients corresponding to the high and low energy data distribution curves of at least two types of calibration objects in the non-reference area, the calibration coefficients corresponding to the high and low energy data distribution curves of other materials not placed on the security inspection machine in the non-reference area are determined by linear interpolation.

[0038] The present invention also provides a calibration device for dual-energy spectral images, comprising:

[0039] The acquisition module is used to acquire multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit.

[0040] The lookup module is used to, for each strip data, look up the target region to which the position in the strip data belongs from a preset calibration coefficient table, and look up the target high-low energy data distribution curve corresponding to the initial high-low energy data in the strip data from multiple candidate high-low energy data distribution curves in the target region, and determine the calibration coefficient corresponding to the target high-low energy data distribution curve as the target calibration coefficient. The calibration coefficient table includes high-low energy data distribution curves of calibrators of different materials in each region, and calibration coefficients corresponding to the high-low energy data distribution curves in each region. The calibration coefficients are used to map the high-low energy data on the corresponding high-low energy data distribution curves to the reference region.

[0041] The calibration module is used to calibrate the initial high and low energy data based on the target calibration coefficient to obtain calibrated high and low energy data.

[0042] The determination module is used to determine the calibration image of the item to be inspected based on the calibrated high and low energy data.

[0043] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the calibration method for dual-energy spectral images as described above.

[0044] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the calibration method for dual-energy spectral images as described above.

[0045] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements a calibration method for dual-energy spectral images as described above.

[0046] The dual-energy spectral image calibration method provided by this invention acquires multiple strip data collected by each detection unit in the detector of a security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit. For each strip data, the target region to which the position in the strip data belongs is found from a preset calibration coefficient table. Then, from multiple candidate high and low energy data distribution curves in the target region, the target high and low energy data distribution curve corresponding to the initial high and low energy data in the strip data is found. The calibration coefficient corresponding to the target high and low energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high and low energy data distribution curves of different materials in each region, as well as calibration coefficients corresponding to the high and low energy data distribution curves in each region. The calibration coefficient is used to map the high and low energy data on the corresponding high and low energy data distribution curve to the reference region. Based on the target calibration coefficient, the initial high and low energy data is calibrated to obtain calibrated high and low energy data. Based on each calibrated high and low energy data, the calibration image of the item to be inspected is determined. Since the strip data of each detection unit can be accurately matched to the corresponding region and the corresponding target high and low energy data distribution curve from the calibration coefficient table according to its position, and then the initial high and low energy data can be mapped to a unified reference region using the corresponding target calibration coefficient, the problem of high and low energy data offset of the same object in different spatial positions can be solved, the color stripes related to pixel position in the calibrated image can be eliminated, and the image quality of the calibrated image can be improved. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0048] Figure 1 A pseudo-color image of a flat plate calibration object.

[0049] Figure 2 This is a schematic diagram of the structure of the X-ray source and detector of the security inspection machine provided in an embodiment of the present invention.

[0050] Figure 3 This is a schematic diagram of the raw high-energy data provided in an embodiment of the present invention.

[0051] Figure 4 This is a schematic diagram of the raw low-energy data provided in an embodiment of the present invention.

[0052] Figure 5 This is a scatter plot of high and low energy data for water provided in an embodiment of the present invention.

[0053] Figure 6 Scatter plot of high and low energy data for alcohol provided in embodiments of the present invention

[0054] Figure 7 This is a schematic flowchart of a dual-energy spectral image calibration method provided in an embodiment of the present invention.

[0055] Figure 8 This is a schematic diagram of the placement of the calibration object provided in an embodiment of the present invention.

[0056] Figure 9 This is a schematic diagram showing the placement of calibration objects of the same material but different thicknesses, provided in an embodiment of the present invention.

[0057] Figure 10 This is a schematic diagram illustrating the division of a calibration object into regions, as provided in an embodiment of the present invention.

[0058] Figure 11 This is a distribution diagram of high and low energy data of calibration objects of different materials in region A, provided in an embodiment of the present invention.

[0059] Figure 12 This is one of the high and low energy data distribution maps of calibration objects of different materials in region B, provided in an embodiment of the present invention.

[0060] Figure 13 This is the second of two calibration high and low energy data distribution diagrams for calibration objects of different materials provided in this embodiment of the invention within region B.

[0061] Figure 14 This is a schematic diagram of the structure of a dual-energy spectral image calibration device provided in an embodiment of the present invention.

[0062] Figure 15 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0064] In the rail transit sector, scanning and inspecting passengers' luggage using X-ray security scanners has become a common practice. These scanners utilize dual-channel high- and low-energy detectors to continuously collect data on the attenuation of X-rays as items pass through the security checkpoint, creating a high- and low-energy data array reflecting the item's density and atomic number. Traditional X-ray image processing algorithms map this high- and low-energy data into a pseudo-color image. Different colors in the pseudo-color image can indicate whether an item is organic, inorganic, or a mixture, but they cannot identify liquids. With technological advancements, it is now possible to utilize the different absorption rates of high- and low-energy X-rays by different liquids. By calculating the absorption ratio, the "equivalent atomic number" of the liquid can be obtained, allowing for the determination of whether the liquid is safe (such as water or beverages), flammable (such as alcohol or gasoline), or chemical (such as pesticides or sulfuric acid). Due to the differences in structure and hardware performance between the X-ray source (X-ray emission module) and detector (data acquisition module) of security inspection machines, the same item placed in different locations will result in significant differences in the high and low energy data collected. Even the "equivalent atomic number" obtained through conversion will deviate, which leads to a decrease in the accuracy of liquid classification.

[0065] Figure 1 This is a pseudo-color image of a flat calibration plate. After placing a uniformly thick, flat calibration plate onto the security inspection machine, images such as... Figure 1 The pseudocolor image shown clearly shows a color change from left to right. The colors in the pseudocolor image represent the material properties of the object, meaning that objects of the same thickness exhibit different material properties in different locations.

[0066] Below, in conjunction with Figures 2-6 Specifically, this explains the issue of high and low energy data shifts for the same item in different spatial locations due to differences in the structure and hardware performance of the X-ray source and detector of the security inspection machine.

[0067] Figure 2 This is a schematic diagram of the structure of the X-ray source and detector of the security inspection machine provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the security inspection machine includes an L-shaped detector with a vertical array and a horizontal array. The X-ray source is located above the detector and emits X-ray beams in a fan shape. The conveyor belt moves in a direction perpendicular to the paper, either inward or outward, that is, perpendicular to the scanning line direction.

[0068] Due to differences in the installation angle and position of different detection units within the detector, the signal performance acquired by different detection units varies significantly. Analysis of the acquired data shows that the high and low energy data collected for liquid identification is highly correlated with position. The detection unit is a detection plate.

[0069] Figure 3This is a schematic diagram of the raw high-energy data provided in an embodiment of the present invention. Figure 4 A schematic diagram of the raw low-energy data provided in the embodiments of the present invention, such as... Figure 3 and Figure 4 As shown, the data collected by each detection unit exhibits distinct bands, even as... Figure 3 and Figure 4 In the image, the right side is mostly an image obtained by capturing the blank area of ​​the conveyor belt, and there will also be stripes with different gray levels.

[0070] Furthermore, Figure 5 This is a scatter plot of high and low energy data for water provided in an embodiment of the present invention. Figure 6 This is a scatter plot of high and low energy data for alcohol provided in an embodiment of the present invention, such as... Figure 5 and Figure 6 As shown, the high-energy and low-energy distribution maps of water and alcohol differ when they are placed at different locations on the conveyor belt. Blue represents the scatter plot at one location, and orange represents the scatter plot at another. The scatter plots show that water and alcohol within the same plastic bottle exhibit shifts in high-energy and low-energy data depending on their spatial location. It should be understood that, ideally... Figure 5 and Figure 6 The scatter plots of blue and orange should overlap.

[0071] In existing technologies, security inspection images are calibrated using background image data, full-scale image data, and original images from the security inspection machine. This calibration reduces the differences in background data between different detectors, enabling the differentiation of organic, inorganic, and mixed substances. However, this calibration method still cannot solve the problem of high-energy data shifts in the same item at different spatial locations. This results in pixel-position-related color banding still appearing in the calibrated image, leading to low image quality.

[0072] In view of the above-mentioned problems, this invention proposes a calibration method for dual-energy spectral images. In this method, after collecting high and low energy data of the item to be inspected in different areas, the high and low energy data of other areas are calibrated to the reference area using a certain area as a reference. In this way, the problem of high and low energy data offset of the same item in different spatial positions can be solved, the color bands related to pixel position in the calibrated image can be eliminated, and the image quality of the calibrated image can be improved.

[0073] The following is combined Figures 7 to 13 The calibration method for dual-energy spectral images provided in the embodiments of the present invention is described below. Here, dual-energy spectral images can be understood as high-energy X-ray images and low-energy X-ray images. The security inspection machines involved in the embodiments of the present invention all refer to security X-ray inspection machines.

[0074] The embodiments of the present invention can be applied to security inspection scenarios in public places such as rail transit, airports and customs, such as subways, high-speed rail and airplanes, to automatically identify whether passengers are carrying prohibited items in their luggage.

[0075] The subject executing this method can be an electronic device such as a security inspection machine, terminal equipment, computer, server, server cluster, or specially designed dual-energy spectrum image calibration device. It can also be a dual-energy spectrum image calibration device installed in the electronic device. The dual-energy spectrum image calibration device can be implemented by software, hardware, or a combination of both.

[0076] Figure 7 This is a schematic flowchart of the dual-energy spectral image calibration method provided in an embodiment of the present invention, as shown below. Figure 7 As shown, the method includes:

[0077] Step 701: Obtain multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit.

[0078] In this step, each detection unit in the detector of the security inspection machine is an independent detection channel that constitutes an L-shaped detector, such as a detector unit with a width of 1mm in the vertical array and the horizontal array.

[0079] As items awaiting security inspection pass through the X-ray beam emitted by the X-ray source of the security scanner on the conveyor belt, each detection unit collects data on the attenuation of X-rays of different energies on the items, thus obtaining strip data collected by each detection unit. The strip data collected by each detection unit includes the position of that unit and the initial high and low energy data collected by that unit. The position of the detection unit is either its detector channel number in the L-shaped array or its horizontal column number within the scan line. Both the detector channel number and the horizontal column number correspond one-to-one with its spatial position along the width of the conveyor belt. By determining the position of the detection unit, the spatial position of the initial high and low energy data collected by that unit along the width of the conveyor belt can be identified.

[0080] Step 702: For each strip data, find the target region to which the position in the strip data belongs from the preset calibration coefficient table, and find the target high and low energy data distribution curve corresponding to the initial high and low energy data in the strip data from multiple candidate high and low energy data distribution curves in the target region, and determine the calibration coefficient corresponding to the target high and low energy data distribution curve as the target calibration coefficient.

[0081] The calibration coefficient table includes high and low energy data distribution curves of calibrators of different materials in each region, as well as calibration coefficients corresponding to the high and low energy data distribution curves in each region. The calibration coefficients are used to map the high and low energy data on the corresponding high and low energy data distribution curves to the reference region.

[0082] In this step, a calibration coefficient table can be generated beforehand through calibration. By placing calibration objects of different materials and thicknesses on the conveyor belt of the security inspection machine and collecting high and low energy data of these calibration objects in different areas, a polynomial fitting method can be used to generate high and low energy data distribution curves for each material in different areas. It should be understood that for the same area, after determining the high and low energy data distribution curves of a finite number of calibration objects of different materials within that area, countless high and low energy data distribution curves within that area can be fitted using interpolation. Therefore, in practical applications, for any pair of initial high and low energy data, the high and low energy data distribution curve corresponding to that initial pair can always be determined in the calibration coefficient table.

[0083] Furthermore, after generating high- and low-energy data distribution curves for calibration objects of various materials in different regions, a certain region is designated as a reference region, such as the central region with the smallest geometric distortion, while other regions are designated as non-reference regions. For calibration objects of the same material, the high- and low-energy data distribution curves corresponding to the calibration objects of that material in a certain non-reference region can be approximated to the high- and low-energy data distribution curves corresponding to the calibration objects of that material in the reference region. This yields the standard coefficients corresponding to the high- and low-energy data distribution curves of the calibration objects of that material in the non-reference region. Therefore, these calibration coefficients can be used to map the high- and low-energy data on the high- and low-energy data distribution curves of the calibration objects of that material in the non-reference region to the calibration region.

[0084] Using the above method, the calibration coefficients corresponding to each high- and low-energy data distribution curve in each non-reference region can be obtained. It should be understood that, for a specific non-reference region, after determining the calibration coefficients corresponding to two high- and low-energy data distribution curves, the calibration coefficients corresponding to other high- and low-energy data distribution curves can be determined through interpolation based on these two calibration coefficients.

[0085] Furthermore, the high and low energy data distribution curves of calibration objects of different materials in each region, along with the corresponding calibration coefficients, can be stored in a calibration coefficient table. In other words, this calibration coefficient table stores the correspondence between regions, high and low energy data distribution curves, and the calibration coefficients corresponding to each high and low energy data distribution curve. For example, region B contains high and low energy data distribution curves CB1, CB2…CBn. The calibration coefficients corresponding to CB1 include k1, c1, r1, and d1; those corresponding to CB2 include k2, c2, r2, and d2; and those corresponding to CBn include kn, cn, rn, and dn. Similarly, region C contains high and low energy data distribution curves CC1, CC2…CCn. The calibration coefficients corresponding to CC1 include k'1, c'1, r'1, and d'1; those corresponding to CC2 include k'2, c'2, r'2, and d'2; and those corresponding to CCn include k'n, c'n, r'n, and d'n. Among them, the calibration coefficients k1, c1, r1 and d1 are used to map the high and low energy data on the high and low energy data distribution curve CB1 to the reference region A.

[0086] In addition, the regions in the calibration coefficient table can be represented by the range of detector channel numbers or the range of horizontal column numbers of the detector unit within the scan line. For example, if the horizontal column number is in the range of 0-120, it is marked as region A, and if the horizontal column number is in the range of 121-250, it is marked as region B, etc.

[0087] By setting high and low energy data distribution curves in different regions and corresponding calibration coefficients in the calibration coefficient table, high and low energy data in other non-reference regions can be mapped to the reference region. This solves the problem of high and low energy data offset of the same item in different spatial locations and eliminates color bands related to pixel position in the calibrated image.

[0088] For each collected strip of data, based on the positions contained within that strip, the spatial location of the high and low energy data along the width of the conveyor belt can be determined. This allows the target region to be found in the calibration coefficient table. For example, if the horizontal column number of a strip is 150, then the target region corresponding to that position is region B.

[0089] In the preset calibration coefficient table, region B contains multiple candidate high and low energy data distribution curves. By comparing the initial high and low energy data with each candidate high and low energy data distribution curve, it can be determined which candidate high and low energy data distribution curve the initial high and low energy data falls on, and thus the target high and low energy data distribution curve corresponding to the initial high and low energy data in the strip data can be determined.

[0090] Furthermore, after determining the target high and low energy data distribution curve, the calibration coefficient corresponding to the target high and low energy data distribution curve is determined as the target calibration coefficient.

[0091] In another possible implementation, for any given region, after determining the calibration coefficients corresponding to any two high- and low-energy data distribution curves through calibration, the calibration coefficients corresponding to all high- and low-energy data point pairs within that region can be determined through interpolation. In this approach, a pre-defined calibration coefficient table stores the calibration coefficients corresponding to all high- and low-energy data point pairs in each region.

[0092] Therefore, for each strip data collected, after determining the target region to which the position belongs based on the position in the strip data from the preset calibration coefficient table, the target calibration coefficient corresponding to the initial high and low energy data is directly determined from the calibration coefficients corresponding to all high and low energy data pairs within the target region.

[0093] Step 703: Based on the target calibration coefficient, calibrate the initial high and low energy data to obtain calibrated high and low energy data.

[0094] In this step, after determining the target calibration coefficient, a linear mapping transformation is performed on the initial high and low energy data based on the target calibration coefficient. This allows the initial high and low energy data to be accurately mapped to the reference region, resulting in calibrated high and low energy data. This eliminates spatial correlation errors caused by differences in detector installation angle and geometric position, and yields calibrated high and low energy data that are independent of position and only reflect the decay characteristics of the material itself.

[0095] Step 704: Based on the calibrated high and low energy data, determine the calibration image of the item to be inspected.

[0096] In this step, all the initial high and low energy data scanned are calibrated. After obtaining each calibrated high and low energy data, they can be spatially rearranged according to the row number of their original scan line and the horizontal column number within the scan line to reconstruct calibrated high and low energy data with dimensions completely consistent with the initial high and low energy data array. Thus, based on the reconstructed high and low energy data, a calibrated image of the item to be inspected can be mapped.

[0097] The dual-energy spectral image calibration method provided in this invention acquires multiple strip data collected by each detection unit in the detector of a security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit. For each strip data, the target region to which the position in the strip data belongs is found from a preset calibration coefficient table. Then, from multiple candidate high and low energy data distribution curves in the target region, the target high and low energy data distribution curve corresponding to the initial high and low energy data in the strip data is found. The calibration coefficient corresponding to the target high and low energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high and low energy data distribution curves of different materials in each region, as well as calibration coefficients corresponding to the high and low energy data distribution curves in each region. The calibration coefficient is used to map the high and low energy data on the corresponding high and low energy data distribution curve to the reference region. Based on the target calibration coefficient, the initial high and low energy data is calibrated to obtain calibrated high and low energy data. Based on each calibrated high and low energy data, the calibration image of the item to be inspected is determined. Since the strip data of each detection unit can be accurately matched to the corresponding region and the corresponding target high and low energy data distribution curve from the calibration coefficient table according to its position, and then the initial high and low energy data can be mapped to a unified reference region using the corresponding target calibration coefficient, the problem of high and low energy data offset of the same object in different spatial positions can be solved, the color stripes related to pixel position in the calibrated image can be eliminated, and the image quality of the calibrated image can be improved.

[0098] For example, based on the above embodiments, the initial high and low energy data can be calibrated using the following formulas (1) and (2) to obtain calibrated high and low energy data:

[0099] (1)

[0100] (2)

[0101] in, This indicates the calibrated high-energy data. This represents the initial high-energy data. This indicates the calibrated low-energy data. The initial low-energy data is represented by k1, c1, r1, and d1, which represent the target calibration coefficients.

[0102] By using the above method, each pair of initial high and low energy data can be calibrated, and the initial high and low energy data in non-reference areas can be uniformly converted to reference areas, so that the high and low energy data of the same substance in different locations tend to be consistent, thereby solving the problem of high and low energy data offset of the same item in different spatial locations.

[0103] The following section provides a detailed explanation of the high and low energy data distribution curves of different materials in each region of the standard coefficient table, as well as the determination method of the calibration coefficients corresponding to the high and low energy data distribution curves in each region.

[0104] For example, when determining the high and low energy data distribution curves of calibration objects of different materials in various regions, it can be done in the following way:

[0105] For calibration objects of different thicknesses and materials placed on the conveyor belt of the security inspection machine, the calibration objects of different thicknesses are divided into multiple regions along a direction perpendicular to the running direction of the conveyor belt. For each region, the high and low energy data of calibration objects of different thicknesses collected by the detection unit in the region are obtained. Based on the high and low energy data of calibration objects of different thicknesses, the high and low energy data distribution curve of the calibration object in the region is determined, and the high and low energy data distribution curve of the calibration objects of different materials in each region is stored in the calibration coefficient table.

[0106] Specifically, flat calibration objects of different thicknesses and materials can be placed on the conveyor belt. Figure 8 This is a schematic diagram of the placement of the calibration object provided in an embodiment of the present invention, such as... Figure 8 As shown, for a calibration object of a certain material and thickness, after being placed flat on a conveyor belt, the security inspection machine can collect high and low energy data of the calibration object in different areas as the conveyor belt moves. In order to acquire high and low energy data of all transverse areas simultaneously in one scan and achieve full-width calibration, the calibration object can be placed along the width of the conveyor belt, and the length of the calibration object is greater than or equal to the width of the conveyor belt.

[0107] Figure 9 This is a schematic diagram illustrating the placement of calibration objects of the same material but different thicknesses, as provided in an embodiment of the present invention. Figure 9 As shown, for calibration objects of the same material but different thicknesses, during calibration, calibration objects of different thicknesses can be placed simultaneously along the width of the conveyor belt, thus allowing for the simultaneous acquisition of high and low energy data for the calibration objects of the same material at different thicknesses. Of course, it can also be done as follows... Figure 8 As shown, only one type of calibration material of a certain thickness is placed at a time, thereby collecting high and low energy data multiple times.

[0108] The materials used for calibration include at least two of the following: iron, copper, aluminum, plexiglass, polyvinyl chloride (PVC), water in plastic containers, and alcohol in plastic containers. By collecting high and low energy data from these calibration materials, a wide range of material types, from metals and organic materials to liquids, can be covered, providing comprehensive calibration data support for the subsequent generation of calibration coefficient tables and accurate classification of liquid substances.

[0109] Because the target is expected Figure 8 The high and low energy data of a certain point P can remain consistent across different security scanners. Since the security scanner uses a line scanning mechanism, when the calibrated object passes through it, it is scanned along lines perpendicular to the conveyor belt's direction of travel, forming the high and low energy data. Therefore, it can be... Figure 8 The calibration objects are divided into multiple regions along a direction perpendicular to the direction of conveyor belt movement. Figure 10 This is a schematic diagram illustrating the division of the calibration object into regions, as provided in an embodiment of the present invention. Figure 10 As shown, the entire calibration object can be evenly divided into N equal parts, i.e., N regions, along the horizontal axis. Figure 10 Taking N=7 as an example. In one possible implementation, a region can be 16 pixels wide or 16 detector units. The smaller the pixel width of each region, the more accurate the calibration result. In the extreme case, each region can degenerate into a point P, that is, in this extreme case, each point is a region.

[0110] For each divided region, for a calibration object of a certain material, the high and low energy data of the calibration object of that material at different thicknesses collected by the detection unit in that region can be obtained. For example, assuming that the calibration object of material C1 has 5 thicknesses, the high and low energy data of the calibration objects of these 5 thicknesses in region A can be obtained respectively.

[0111] When acquiring calibration high and low energy data of calibration objects of different thicknesses collected by the detection units in each region, the original calibration high and low energy data of each calibration object of different thicknesses collected by the detection units in the region can be acquired, and full-scale calibration can be performed on each original calibration high and low energy data to obtain calibration high and low energy data.

[0112] Specifically, full-scale calibration involves laying a uniform standard absorber (such as a block of plexiglass or aluminum of a specific thickness) flat on a conveyor belt and allowing it to pass completely through the X-ray beam. The maximum count values ​​of the high and low energy channels of each detector unit under full-load X-ray penetration are then recorded to establish a full-scale response benchmark. Subsequently, when collecting the original high and low energy data of the calibration material, the count value measured by each detector unit is subtracted from the baseline value of that detector unit, and then divided by the full-scale benchmark value of that detector unit to complete the normalization mapping. This eliminates the dimensional inconsistency caused by differences in detector gain and yields calibration high and low energy data that only reflect the decay characteristics of the material.

[0113] In the specific implementation process, multiple calibration high and low energy data will be obtained for points in the same area or on the scan line at a certain thickness. The high and low energy data obtained at the center point of this area can be used as the calibration high and low energy data for this area at that thickness. Alternatively, the high and low energy data of a certain set point P can be used as the calibration high and low energy data for this area at that thickness. Alternatively, the average value of all high energy data and the average value of all low energy data obtained in this area can be determined, and the resulting average value can be used as the calibration high and low energy data for this area at that thickness. Alternatively, the calibration high and low energy data for this area at that thickness can be determined by fitting.

[0114] Figure 11 This is a distribution map of high and low energy data of calibration objects of different materials in region A provided by an embodiment of the present invention, such as... Figure 11 As shown, with the calibration high-energy data H as the x-axis and the calibration low-energy data L as the y-axis, the five yellow dots represent the calibration high and low energy data of the calibration object (a flat calibration plate) at point P on the scan line for five different thicknesses of material C1 within region A. Each yellow dot represents the calibration high and low energy data collected at one thickness. The white diamond-shaped dots represent the calibration high and low energy data of the calibration object at point P for five different thicknesses of material C2 within region A.

[0115] Figure 12 This is one of the high and low energy data distribution maps of calibration objects of different materials in region B provided in this embodiment of the invention, such as... Figure 12 As shown, data collection was performed in the same manner within region B to obtain calibration high and low energy data for calibration objects of different materials and thicknesses within region B. The high-energy calibration data H is plotted on the x-axis, and the low-energy calibration data L is plotted on the y-axis. Five green dots represent the calibration high and low energy data at point P on the scan line for calibration objects (flat calibration plates) of material C1 at five different thicknesses within region B. Each green dot represents the calibration high and low energy data collected at one thickness. Blue diamond-shaped dots represent the calibration high and low energy data at point P for calibration objects of material C2 at five different thicknesses within region B.

[0116] For a calibration object of a certain material, after obtaining the calibration high and low energy data at different thicknesses in a certain region, the distribution curve of the high and low energy data of the calibration object of that material in this region can be determined by polynomial fitting.

[0117] Using the above method, the high and low energy data distribution curves of calibration materials of different materials in each region can be determined, and the high and low energy data distribution curves of calibration materials of different materials in each region can be stored in the calibration coefficient table.

[0118] In this embodiment, by acquiring calibration high and low energy data of calibration objects of different materials at different thicknesses in various regions, and based on the calibration high and low energy data at different thicknesses, the high and low energy data distribution curves of calibration objects of different materials in different regions are fitted, so that the fitted high and low energy data distribution curves can be adapted to items of different thicknesses. When performing image calibration in the future, the corresponding target high and low energy data distribution curves can be accurately found in the calibration coefficient table.

[0119] For example, based on the above embodiments, when determining the high and low energy data distribution curve of the calibration object in a region based on the calibration high and low energy data of calibration objects with different thicknesses, a polynomial fitting algorithm can be used for calibration objects of various materials to perform polynomial fitting on the calibration high and low energy data of the calibration object at different thicknesses in the region to obtain the high and low energy data distribution curve of the calibration object in the region.

[0120] Specifically, for a calibration object of the same material in a certain region, after obtaining the calibration high and low energy data of the calibration object of different thicknesses in this region, a polynomial curve fitting algorithm can be used to determine the distribution curve of the high and low energy data of the calibration object in this region. Among them, the core of the polynomial curve fitting algorithm is to approximate a set of discrete data points with a polynomial function so that the error (usually the sum of squares) of the polynomial at these points is minimized. The essence of polynomial curve fitting is to solve an overdetermined system of equations and find the optimal polynomial coefficients by the least squares method. Assume that the fitted polynomial is represented by the following formula (4):

[0121] (4)

[0122] in, denoted by h, which indicates calibration of low-energy data; denoted by h, which indicates calibration of high-energy data; and denoted by n, which indicates the degree of the polynomial, typically set to 3. This represents the coefficients to be determined in the polynomial.

[0123] For calibration materials of different thicknesses of the same material, m calibration high and low energy data in a certain region. The error function is defined as shown in the following formula (5):

[0124] (5)

[0125] Find the answer for E respectively Taking the partial derivatives and setting them to zero, we obtain a system of linear equations. Solving this system of equations yields the coefficients. .

[0126] After obtaining the coefficients using the above method, the high and low energy data distribution curve of the calibrator in a certain region can be obtained.

[0127] In this embodiment, a polynomial fitting algorithm can be used to perform polynomial fitting on the calibration high and low energy data of the calibration object at different thicknesses within the region, thereby obtaining the high and low energy data distribution curve of the calibration object within the region. This allows the fitted high and low energy data distribution curve to provide corresponding high and low energy data at any thickness. When calibrating the image of the item to be inspected in the subsequent process, even if the thickness of the item to be inspected is inconsistent with the thickness of the calibration object, the corresponding target high and low energy data distribution curve can be accurately found, ensuring accurate correction of the images of items to be inspected at different thicknesses.

[0128] For example, when determining the calibration coefficients corresponding to each high- and low-energy data distribution curve, it can be done in the following way:

[0129] A reference region is determined from multiple regions. For any non-reference region, the high and low energy data distribution curve of a calibrator of any material in the non-reference region is approximated to the high and low energy data distribution curve of the same material in the reference region through a linear transformation. The calibration coefficients corresponding to the high and low energy data distribution curves of the calibrators of different materials in each non-reference region are obtained and stored in the calibration coefficient table.

[0130] Specifically, the reference area is a pre-selected reference area among multiple areas within the horizontal scanning range. This reference area can be any one of the areas or a geometrically centered area.

[0131] For any other non-reference region, a linear transformation can be used to establish a parameter mapping relationship, so that the high and low energy data distribution curve of a calibration object of a certain material in the non-reference region is as close as possible to the high and low energy data distribution curve of the calibration object of the same material in the reference region, thereby obtaining the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of this material in the non-reference region.

[0132] For example, assuming the reference area is region A and the non-reference area is region B, the high and low energy data distribution curve of the calibration material C1 in region A is CA1, and the high and low energy data distribution curve in region B is CB1. By establishing a parameter mapping relationship, the high and low energy data distribution curve CB1 is made as close as possible to the high and low energy data distribution curve CA1, thereby obtaining the calibration coefficient corresponding to the high and low energy data distribution curve CB1.

[0133] By using the above method, the calibration coefficients corresponding to the high and low energy data distribution curves of all materials in each non-reference area can be obtained, and the correspondence between the area, the high and low energy distribution data curves and the calibration coefficients can be stored in the calibration coefficient table. This provides a basis for the accurate lookup of the target calibration coefficients in the future.

[0134] For example, when using linear transformation to approximate the high and low energy data distribution curve of a calibration object of any material in a non-reference region to the high and low energy data distribution curve of a calibration object of the same material in a reference region, and obtaining the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of the material in the non-reference region, the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of the material in the non-reference region can be determined according to formula (3):

[0135] (3)

[0136] in, , representing the polynomial of the high-energy data distribution curve of the calibration material of the same material within the reference area. This represents the high-energy data of the calibration material within the reference area. The polynomial coefficients represent the high-energy data distribution curves of calibration materials of the same material within the reference region. , The polynomial coefficients represent the high and low energy data distribution curves of the calibration material in the non-reference area, where k, c, r, and d represent calibration coefficients.

[0137] Specifically, assuming the reference region is region A, for the non-reference region B, the curves of the calibration objects of the same material in region B and region A need to be as close as possible, or even overlap. Therefore, approximation can be achieved by establishing a parameter mapping relationship. Assuming the curves in region A... For the target curve, let the curve of the calibration object of the same material in region B be... to approach For the target curve and the curve to be approximated Establish parameter mapping Define the error function: ,in This represents the maximum value of the high-energy parameter. The parameter mapping is solved by minimizing the error function. .

[0138] Since a polynomial curve fitting is used, and the polynomial curves of region A and region B are of the same order, the approximation process can be simplified. A combination of linear transformations on the horizontal axis h (high-energy data) and the vertical axis l (low-energy data) can be used to achieve accurate mapping: target curve and the curve to be approximated The polynomials have the same degree, both being nth degree polynomials. Given the complete analytical expressions of two curves:

[0139]

[0140]

[0141] in, and , The polynomial coefficients represent the high-energy data distribution curve of a calibration material of a certain material within the reference region A. The polynomial coefficients represent the distribution curves of high and low energy data for the same material in non-reference region B.

[0142] A linear combination of horizontal axis h-transformation and vertical axis l-transformation is used, with horizontal axis h-transformation plus scaling: Where k is the scaling factor for the horizontal axis h, c is the translation factor for the horizontal axis h, and the vertical axis l is transformed and scaled: Where r is the scaling factor of the vertical axis l, and d is the translation factor of the vertical axis l. Therefore, the ultimate goal is to make ,Right now ,Will and Substituting the parameters of the curve's analytical expression into the above equation, and solving it, we can obtain the mapping parameters k, c, r, and d. It should be understood that... To be The parameter h in the analytical expression of the curve is replaced with That is, .

[0143] In this embodiment, by establishing a parameter mapping method, the calibration coefficients corresponding to the high and low energy data distribution curves of each material in the non-reference area can be determined. Thus, the high and low energy data in the non-reference area can be mapped to the reference area through the calibration coefficients. This can eliminate spatial correlation errors caused by differences in detector installation angle and geometric position, and make the high and low energy data of the same material in different lateral positions more consistent, thereby improving the accuracy of subsequent image calibration.

[0144] For example, based on the above embodiments, the calibration coefficients corresponding to the high and low energy data distribution curves of calibration objects of at least two materials in the non-reference area can also be determined by linear interpolation based on the calibration coefficients corresponding to the high and low energy data distribution curves of calibration objects of at least two materials in the non-reference area.

[0145] Specifically, Figure 13 This is the second example of the high and low energy data distribution diagrams of calibration objects of different materials in region B provided by the embodiments of the present invention. Figure 13As shown, through the approximation mapping method described in the aforementioned embodiments, the calibration coefficients k1, c1, r1, and d1 from the high- and low-energy data distribution curve CB1 to the curve CA1 can be obtained. Similarly, the mapping parameters k2, c2, r2, and d2 from the curve CB2 to the curve CA2 can also be approximated. Since in practical applications, the materials of items to be inspected are numerous, and it is impossible to place all material calibration objects on the security inspection machine for calibration, after determining the high- and low-energy data distribution curves of at least two material calibration objects, a curve corresponding to any other material calibration object not placed on the security inspection machine can be fitted using linear interpolation. In an extreme case, the fitted high- and low-energy data distribution curve will be filled with... Figure 13 The entire HL plane shown can generate high and low energy data distribution curves for any uncalibrated material in any region by calibrating and interpolating only two known materials, thereby achieving full material and full region coverage with minimal calibration cost.

[0146] In one possible implementation, for any fitted high- and low-energy data distribution curve in the non-reference region B, such as Figure 13 The calibration coefficients corresponding to the high- and low-energy data distribution curve CB3 can be determined by linear interpolation using calibration coefficients k1, c1, r1, d1 and k2, c2, r2, d2. Using this method, the calibration coefficients from any high- and low-energy data distribution curve in non-reference region B to the reference region A can be determined. Alternatively, the calibration coefficients from any pair of high- and low-energy data points in non-reference region B to the reference region A can also be determined using the same linear interpolation method.

[0147] In another implementation, for any high- and low-energy data distribution curve fitted within the non-reference region B, a linear transformation can be used to establish a parameter mapping, approximating the fitted high- and low-energy data distribution curve of the same material within the reference region. This yields the calibration coefficients corresponding to the high- and low-energy data distribution curves within the non-reference region B. It should be noted that in this method, the high- and low-energy data distribution curves of the same material within the reference region are also obtained through fitting.

[0148] In this embodiment, by using the calibration coefficients corresponding to the high and low energy data distribution curves of at least two materials in the non-reference area, and through linear interpolation, the calibration coefficients corresponding to the high and low energy data distribution curves of other materials not placed on the security inspection machine in the non-reference area can be determined. This allows for full material and full area calibration coefficient coverage with minimal calibration workload, reducing calibration costs.

[0149] Furthermore, in this invention, after calibrating the initial high and low energy data, the material properties of the high and low energy images at different spatial locations are consistent. When identifying the types of substances based on the calibrated high and low energy data, it is possible not only to accurately distinguish between organic and inorganic substances, but also to accurately determine the specific type of liquid through the "equivalent atomic number".

[0150] In addition, after determining the calibration image of the item to be inspected based on the high and low energy data after each calibration, the calibration image can be used not only for the identification of the type of substance, but also for model training, and for quality assessment of the security inspection image, etc.

[0151] The calibration apparatus for dual-energy spectral images provided by the present invention will be described below. The calibration apparatus for dual-energy spectral images described below can be referred to in correspondence with the calibration method for dual-energy spectral images described above.

[0152] Figure 14 This is a schematic diagram of the structure of the calibration device for dual-energy spectral images provided in an embodiment of the present invention, as shown below. Figure 14 As shown, the calibration device 1400 for the dual-energy spectral image includes:

[0153] The acquisition module 11 is used to acquire multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit.

[0154] The lookup module 12 is used to look up the target region to which the position in the strip data belongs from a preset calibration coefficient table for each strip data, and to look up the target high-low energy data distribution curve corresponding to the initial high-low energy data in the strip data from multiple candidate high-low energy data distribution curves in the target region, and to determine the calibration coefficient corresponding to the target high-low energy data distribution curve as the target calibration coefficient. The calibration coefficient table includes high-low energy data distribution curves of calibrators of different materials in each region, and calibration coefficients corresponding to the high-low energy data distribution curves in each region. The calibration coefficients are used to map the high-low energy data on the corresponding high-low energy data distribution curves to the reference region.

[0155] The calibration module 13 is used to calibrate the initial high and low energy data based on the target calibration coefficient to obtain calibrated high and low energy data.

[0156] The determination module 14 is used to determine the calibration image of the item to be inspected based on the calibrated high and low energy data.

[0157] In one example embodiment, calibration module 13 is specifically used for:

[0158] Based on the following formulas (1) and (2), the initial high and low energy data are calibrated to obtain the calibrated high and low energy data:

[0159] (1)

[0160] (2)

[0161] in, This indicates the calibrated high-energy data. This represents the initial high-energy data. This indicates the calibrated low-energy data. The initial low-energy data is represented by k1, c1, r1, and d1, which represent the target calibration coefficients.

[0162] In one example embodiment, the device further includes:

[0163] The segmentation module is used to divide the calibration objects of different thicknesses and materials placed on the conveyor belt of the security inspection machine into multiple areas along a direction perpendicular to the running direction of the conveyor belt.

[0164] The acquisition module 11 is also used to acquire calibration high and low energy data of calibration objects of different thicknesses collected by the detection unit in each region;

[0165] The determination module 14 is also used to determine the high and low energy data distribution curve of the calibration material in the region based on the calibration high and low energy data of the calibration material with different thicknesses;

[0166] The storage module is used to store the high and low energy data distribution curves of the calibration materials of different materials in each region in the calibration coefficient table.

[0167] In one example embodiment, the determining module 14 is specifically used for:

[0168] For each material of the calibration object, a polynomial fitting algorithm is used to perform polynomial fitting on the calibration high and low energy data of the calibration object at different thicknesses in the region, so as to obtain the high and low energy data distribution curve of the calibration object in the region.

[0169] In one example embodiment, the acquisition module 11 is specifically used for:

[0170] Obtain the original calibration high and low energy data of calibration objects of different thicknesses collected by the detection unit within the region;

[0171] Full-scale calibration is performed on each of the original calibration high and low energy data to obtain the calibration high and low energy data.

[0172] In one example embodiment, the calibrator is placed along the width direction of the conveyor belt, and the length of the calibrator is greater than or equal to the width of the conveyor belt.

[0173] In one example embodiment, the calibrator includes at least two of the following: iron, copper, aluminum, plexiglass, PVC, water in plastic containers, and alcohol in plastic containers.

[0174] In one example embodiment, the device further includes:

[0175] The determining module 14 is further configured to determine the reference region from the plurality of regions;

[0176] The approximation module is used to approximate the high and low energy data distribution curve of a calibration object of any material in any non-reference region to the high and low energy data distribution curve of a calibration object of the same material in the reference region through linear transformation, thereby obtaining the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of the material in the non-reference region.

[0177] The storage module is also used to store the calibration coefficients corresponding to the high and low energy data distribution curves of different materials in each of the non-reference areas into the calibration coefficient table.

[0178] In one example embodiment, the approximation module is specifically used for:

[0179] According to formula (3), determine the calibration coefficient corresponding to the high and low energy data distribution curve of the material in the non-reference area:

[0180] (3)

[0181] in, , representing the polynomial of the high-energy data distribution curve of the calibration material of the same material within the reference region. This represents the high-energy data of the calibrators within the reference area. The polynomial coefficients represent the high-energy data distribution curves of calibrators of the same material within the reference region. , The polynomial coefficients represent the high and low energy data distribution curves of the material in the non-reference region, where k, c, r, and d represent calibration coefficients.

[0182] In one example embodiment, the determining module 14 is further configured to determine, by linear interpolation, the calibration coefficients corresponding to the high and low energy data distribution curves of calibration objects of other materials not placed on the security inspection machine in the non-reference area, based on the calibration coefficients corresponding to the high and low energy data distribution curves of at least two types of calibration objects in the non-reference area.

[0183] The apparatus of this embodiment can be used in any of the methods in the calibration method side embodiment of dual-energy spectral images. Its specific implementation process and technical effects are similar to those in the calibration method side embodiment of dual-energy spectral images. For details, please refer to the detailed description in the calibration method side embodiment of dual-energy spectral images, which will not be repeated here.

[0184] Figure 15 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as... Figure 15 As shown, the electronic device may include: a processor 1510, a communication interface 1520, a memory 1530, and a communication bus 1540, wherein the processor 1510, the communication interface 1520, and the memory 1530 communicate with each other through the communication bus 1540. The processor 1510 can call logical instructions in the memory 1530 to execute a dual-energy spectrum image calibration method. The method includes: acquiring multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected, wherein each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit; for each strip data, searching for the target region to which the position in the strip data belongs from a preset calibration coefficient table, and searching for the target high and low energy corresponding to the initial high and low energy data in the strip data from multiple candidate high and low energy data distribution curves in the target region. The high and low energy data distribution curves are used to determine the calibration coefficients corresponding to the target high and low energy data distribution curves as target calibration coefficients. The calibration coefficient table includes high and low energy data distribution curves of different materials in each region, as well as calibration coefficients corresponding to the high and low energy data distribution curves in each region. The calibration coefficients are used to map the high and low energy data on the corresponding high and low energy data distribution curves to the reference region. Based on the target calibration coefficients, the initial high and low energy data are calibrated to obtain calibrated high and low energy data. Based on each of the calibrated high and low energy data, the calibration image of the item to be inspected is determined.

[0185] Furthermore, the logical instructions in the aforementioned memory 1530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0186] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the dual-energy spectrum image calibration method provided by the above methods. The method includes: acquiring multiple strip data collected by each detection unit in the detector of a security inspection machine for items to be inspected, each strip data including the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit; for each strip data, searching from a preset calibration coefficient table for the target region to which the position in the strip data belongs, and selecting from multiple candidate high and low energy data in the target region. In the data distribution curve, the target high- and low-energy data distribution curve corresponding to the initial high- and low-energy data in the strip data is found, and the calibration coefficient corresponding to the target high- and low-energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high- and low-energy data distribution curves of calibration objects of different materials in each region, as well as calibration coefficients corresponding to the high- and low-energy data distribution curves in each region. The calibration coefficients are used to map the high- and low-energy data on the corresponding high- and low-energy data distribution curves to the reference region. Based on the target calibration coefficient, the initial high- and low-energy data is calibrated to obtain calibrated high- and low-energy data. Based on each of the calibrated high- and low-energy data, the calibration image of the item to be inspected is determined.

[0187] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a calibration method for dual-energy spectral images provided by the methods described above. This method includes: acquiring multiple strip data collected by each detection unit in a security inspection machine for an item to be inspected, each strip data including the position of the corresponding detection unit and initial high and low energy data collected by the detection unit; for each strip data, searching for a target region to which the position in the strip data belongs from a preset calibration coefficient table, and searching for the strip from multiple candidate high and low energy data distribution curves in the target region. The initial high- and low-energy data in the data corresponds to a target high- and low-energy data distribution curve, and the calibration coefficient corresponding to the target high- and low-energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high- and low-energy data distribution curves of calibration objects of different materials in each region, and calibration coefficients corresponding to the high- and low-energy data distribution curves in each region. The calibration coefficients are used to map the high- and low-energy data on the corresponding high- and low-energy data distribution curves to the reference region. Based on the target calibration coefficient, the initial high- and low-energy data is calibrated to obtain calibrated high- and low-energy data. Based on each of the calibrated high- and low-energy data, the calibration image of the item to be inspected is determined.

[0188] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0189] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0190] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A calibration method for dual-energy spectral images, characterized in that, include: The system acquires multiple strip data collected by each detection unit in the detector of the security inspection machine for the item to be inspected. Each strip data includes the position of the corresponding detection unit and the initial high and low energy data collected by the detection unit. For each strip data, the target region to which the position in the strip data belongs is found from a preset calibration coefficient table, and the target high-low energy data distribution curve corresponding to the initial high-low energy data in the strip data is found from multiple candidate high-low energy data distribution curves in the target region. The calibration coefficient corresponding to the target high-low energy data distribution curve is determined as the target calibration coefficient. The calibration coefficient table includes high-low energy data distribution curves of calibrators of different materials in each region, and calibration coefficients corresponding to the high-low energy data distribution curves in each region. The calibration coefficients are used to map the high-low energy data on the corresponding high-low energy data distribution curves to the reference region. Based on the target calibration coefficient, the initial high and low energy data are calibrated to obtain calibrated high and low energy data; Based on the calibrated high and low energy data, the calibration image of the item to be inspected is determined.

2. The calibration method for dual-energy spectral images according to claim 1, characterized in that, The step of calibrating the initial high- and low-energy data based on the target calibration coefficient to obtain calibrated high- and low-energy data includes: Based on the following formulas (1) and (2), the initial high and low energy data are calibrated to obtain the calibrated high and low energy data: （1） （2） in, This indicates the calibrated high-energy data. This represents the initial high-energy data. This indicates the calibrated low-energy data. The initial low-energy data is represented by k1, c1, r1, and d1, which represent the target calibration coefficients.

3. The calibration method for dual-energy spectral images according to claim 1 or 2, characterized in that, The method further includes: For calibration objects of different thicknesses and materials placed on the conveyor belt of the security inspection machine, the calibration objects of different thicknesses are divided into multiple areas along a direction perpendicular to the running direction of the conveyor belt; For each region, high and low energy calibration data of calibration objects of different thicknesses collected by the detection units within the region are obtained; Based on the high and low energy data of the calibration material with different thicknesses, the distribution curve of the high and low energy data of the calibration material in the region is determined; The high and low energy data distribution curves of the calibration materials of different materials in each region are stored in the calibration coefficient table.

4. The calibration method for dual-energy spectral images according to claim 3, characterized in that, The determination of the high and low energy data distribution curve of the calibration material within the region based on calibration data of calibration materials of different thicknesses includes: For each material of the calibration object, a polynomial fitting algorithm is used to perform polynomial fitting on the calibration high and low energy data of the calibration object at different thicknesses in the region, so as to obtain the high and low energy data distribution curve of the calibration object in the region.

5. The calibration method for dual-energy spectral images according to claim 3, characterized in that, The acquisition of calibration high and low energy data of calibration objects of different thicknesses collected by the detection unit within the region includes: Obtain the original calibration high and low energy data of calibration objects of different thicknesses collected by the detection unit within the region; Full-scale calibration is performed on each of the original calibration high and low energy data to obtain the calibration high and low energy data.

6. The calibration method for dual-energy spectral images according to claim 3, characterized in that, The calibration object is placed along the width of the conveyor belt, and the length of the calibration object is greater than or equal to the width of the conveyor belt.

7. The calibration method for dual-energy spectral images according to claim 3, characterized in that, The calibrator includes at least two of the following: iron, copper, aluminum, plexiglass, PVC, water in plastic containers, and alcohol in plastic containers.

8. The calibration method for dual-energy spectral images according to claim 3, characterized in that, The method further includes: The reference region is determined from the plurality of regions; For any non-reference region, the high and low energy data distribution curve of the calibration object of any material in the non-reference region is approximated by the high and low energy data distribution curve of the calibration object of the same material in the reference region through linear transformation, so as to obtain the calibration coefficient corresponding to the high and low energy data distribution curve of the calibration object of the material in the non-reference region. The calibration coefficients corresponding to the high and low energy data distribution curves of different materials in each of the non-reference regions are stored in the calibration coefficient table.

9. The calibration method for dual-energy spectral images according to claim 8, characterized in that, The method of approximating the high- and low-energy data distribution curves of any material in the non-reference region to the high- and low-energy data distribution curves of the same material in the reference region through linear transformation, thereby obtaining the calibration coefficients corresponding to the high- and low-energy data distribution curves of the material in the non-reference region, includes: According to formula (3), determine the calibration coefficient corresponding to the high and low energy data distribution curve of the material in the non-reference area: （3） in, , representing the polynomial of the high-energy data distribution curve of the calibration material of the same material within the reference region. This represents the high-energy data of the calibrators within the reference area. The polynomial coefficients represent the high-energy data distribution curves of calibrators of the same material within the reference region. , The polynomial coefficients represent the high and low energy data distribution curves of the material in the non-reference region, where k, c, r, and d represent calibration coefficients.

10. The calibration method for dual-energy spectral images according to claim 8, characterized in that, The method further includes: Based on the calibration coefficients corresponding to the high and low energy data distribution curves of at least two types of calibration objects in the non-reference area, the calibration coefficients corresponding to the high and low energy data distribution curves of other materials not placed on the security inspection machine in the non-reference area are determined by linear interpolation.

Images