Odct image reconstruction method, apparatus, and program product

Abstract

The application discloses an ODCT image reconstruction method, device and program product, and applies to the technical field of image reconstruction. The method comprises the following steps: orthogonally linearly scanning a target object along a horizontal axis and a vertical axis to obtain first projection data and second projection data of the target object; wherein the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis; performing the operations of weighting, filtering and back projection on the first projection data and the second projection data to obtain back projection data of the first projection data and back projection data of the second projection data; and performing a fusion operation on the back projection data of the first projection data and the back projection data of the second projection data to obtain reconstructed image data of the target object. The application can effectively improve the reconstruction efficiency of the ODCT image while ensuring the reconstruction quality of the ODCT image.

Description

Technical Field

[0001] This application relates to the field of image reconstruction technology, and in particular to an ODCT image reconstruction method, apparatus and program product. Background Technology

[0002] Orthogonal Directional-translation Computed Tomography (ODCT) is essentially a finite-angle CT technique. It acquires truncated projection data and reconstructs the three-dimensional structure of components using finite-angle CT reconstruction algorithms. For finite-angle reconstruction problems, the projection data often fails to meet the data integrity requirements of analytical algorithms. Therefore, iterative algorithms, which have lower data integrity requirements, are more commonly used in effective-angle CT reconstruction. Iterative algorithms require significantly more storage space and have a much longer computation time than analytical algorithms. Thus, most ODCT imaging currently uses iterative algorithms, sacrificing storage space and runtime for high-quality reconstructed images. Summary of the Invention

[0003] This application provides an ODCT image reconstruction method, apparatus, and program product, which effectively improves the reconstruction efficiency of ODCT images while ensuring the reconstruction quality of ODCT images.

[0004] On the one hand, embodiments of this application provide an ODCT image reconstruction method, including the following steps:

[0005] The target object is scanned along the horizontal and vertical axes using orthogonal straight lines to obtain first projection data and second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis.

[0006] The first projection data and the second projection data are weighted, filtered and back-projected to obtain the back-projected data of the first projection data and the back-projected data of the second projection data.

[0007] The back-projection data of the first projection data and the back-projection data of the second projection data are fused to obtain the reconstructed image data of the target object.

[0008] Further, in one embodiment, the operation of weighting, filtering, and backprojecting the first projection data and the second projection data to obtain the backprojected data of the first projection data and the backprojected data of the second projection data includes:

[0009] The target projection data is weighted to obtain the weighted target projection data.

[0010] The weighted target projection data is filtered to obtain the filtered target projection data.

[0011] Perform a back-projection operation on the filtered target projection data to obtain the back-projected data of the target projection data;

[0012] The target projection data is either the first projection data or the second projection data.

[0013] Further, in one embodiment, the weighting operation on the target projection data to obtain the weighted target projection data includes:

[0014] The weighting coefficients of the target projection data are calculated based on the scanning parameters associated with orthogonal linear scanning.

[0015] The target projection data is weighted based on the weighting coefficients to obtain the weighted target projection data.

[0016] Further, in one embodiment, the scanning parameters include the movement distance of the X-ray source, the physical distance of the detector in a first direction, the physical distance of the detector in a second direction, the distance from the plane in which the X-ray source moves to the center plane of the target object, and the distance from the plane in which the X-ray source moves to the plane in which the detector moves, wherein the first direction is a direction parallel to the orthogonal linear scanning trajectory, and the second direction is a direction perpendicular to the orthogonal linear scanning trajectory.

[0017] The step of calculating the weighting coefficients of the target projection data based on the scanning parameters related to orthogonal linear scanning includes:

[0018] The cosine value of the cone angle is calculated based on the distance from the plane of motion of the radiation source to the plane of motion of the detector and the physical distance of the detector in the second direction.

[0019] The square root operation is performed based on the cosine value of the cone angle, the distance from the plane where the ray source is moving to the center plane of the target object, the distance from the plane where the ray source is moving to the plane where the detector is moving, the movement distance of the ray source, and the physical distance of the detector in the first direction to obtain the square root operation value;

[0020] The weighting coefficient is determined by the ratio of the distance from the plane where the ray source is moving to the center plane of the target object to the square root value.

[0021] Further, in one embodiment, the step of filtering the weighted target projection data to obtain filtered target projection data includes:

[0022] Construct a ramp filter for the target projection data;

[0023] Perform a Fast Fourier Transform on the weighted target projection data to obtain the transformed target projection data;

[0024] The transformed target projection data is processed using the ramp filter to obtain the processed target projection data;

[0025] The processed target projection data is subjected to inverse Fourier transform to obtain filtered target projection data.

[0026] Further, in one embodiment, the step of performing a back-projection operation on the filtered target projection data to obtain the back-projection data of the target projection data includes:

[0027] Based on the ordinate of the reconstructed point in the target projection data in the reconstruction coordinate system, combined with the distance from the plane where the ray source is moving to the center plane of the target object, and the distance from the plane where the ray source is moving to the plane where the detector is moving, the distance factor of the reconstructed point in the target projection data on the ray is obtained.

[0028] Using the distance factor, a back-projection operation is performed on the filtered target projection data to obtain the back-projected data of the target projection data.

[0029] Further, in one embodiment, the step of obtaining the distance factor of the reconstructed point on the ray in the target projection data based on the ordinate of the reconstructed point in the reconstructed coordinate system, combined with the distance from the plane where the ray source moves to the center plane of the target object, and the distance from the plane where the ray source moves to the plane where the detector moves, includes:

[0030] The distance from the plane where the ray source is moving to the center plane of the target object and the sum of the ordinates of the reconstructed points in the target projection data in the reconstructed coordinate system are obtained as the target distance;

[0031] The distance factor is defined as the ratio of the target distance to the distance between the plane of motion of the radiation source and the plane of motion of the detector.

[0032] Further, in one embodiment, the step of fusing the back-projection data of the first projection data and the back-projection data of the second projection data to obtain the reconstructed image data of the target object includes:

[0033] The sum of the backprojection data of the first projection data and the backprojection data of the second projection data is determined as the reconstructed image data.

[0034] On the other hand, embodiments of this application provide an ODCT image reconstruction apparatus, comprising:

[0035] The first processing module is used to perform orthogonal linear scanning of the target object along the horizontal axis and the vertical axis to obtain the first projection data and the second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis.

[0036] The second processing module is used to perform weighting, filtering and back-projection operations on the first projection data and the second projection data to obtain the back-projection data of the first projection data and the back-projection data of the second projection data.

[0037] The third processing module is used to perform a fusion operation on the back-projection data of the first projection data and the back-projection data of the second projection data to obtain the reconstructed image data of the target object.

[0038] In another aspect, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the aforementioned ODCT image reconstruction method.

[0039] According to an embodiment of this application, an ODCT image reconstruction method, apparatus, and program product are provided. An orthogonal linear scan of a target object is performed along a horizontal and vertical axis to obtain first and second projection data of the target object. The first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis. Weighting, filtering, and backprojection operations are performed on the first and second projection data to obtain backprojection data of the first and second projection data. The backprojection data of the first and second projection data are then fused to obtain reconstructed image data of the target object. This application performs weighting, filtering, backprojection, and fusion of two orthogonal linear scan projection data through a unified projection domain coordinate mapping to perform local reconstruction of the target object, thereby effectively improving the reconstruction efficiency of ODCT images while ensuring the reconstruction quality. Attached Figure Description

[0040] Figure 1 This is a flowchart of an ODCT image reconstruction method provided in this application;

[0041] Figure 2 This is a schematic diagram of the geometric optical path and motion path corresponding to the data acquisition stage provided in this application;

[0042] Figure 3 This is an example diagram of the mechanical structure for acquiring ODCT projection data provided in this application;

[0043] Figure 4 This is a schematic diagram of the geometric optical path and motion path for acquiring ODCT projection data provided in this application;

[0044] Figure 5 This is the reconstruction result image provided in this application after the data backprojection and fusion stage;

[0045] Figure 6 This is a structural diagram of an ODCT image reconstruction device provided in this application. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0047] The present application will be further described below with reference to the accompanying drawings and specific embodiments. The described embodiments should not be considered as limitations on the present application, and all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present application.

[0048] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0050] First, the terms and nouns used in the embodiments of this application will be explained and described:

[0051] (1) Computed Tomography (CT) is an imaging technology that uses the principle of X-ray transmission to obtain information about the three-dimensional structure inside an object.

[0052] (2) Computed Laminography (CL) is an imaging technique mainly used for the detection of plate-shaped components. It can avoid the long path direction of plate-shaped objects by using tilt scanning or limited angle.

[0053] (3) Orthogonal linear scanning refers to the two mutually perpendicular linear translation movements of the object relative to the imaging system, rather than the traditional object rotation or X-ray source rotation scanning.

[0054] (4) Orthogonal Directional-translation Computed Tomography (ODCT) is a derivative CT technology. In this technology, the X-ray source and the detector are parallel to each other and move in a straight line to scan the object to be detected along two mutually orthogonal straight lines. The projection data of the object to be detected in the two mutually orthogonal directions are obtained, and the image is reconstructed accordingly. It has the advantage of achieving high-resolution imaging while maintaining the original high imaging speed.

[0055] (5) Orthogonal Translation Computed Laminography (OTCL) is a derivative CL technology. It can also be regarded as one of the derivative technologies of ODCT. It performs imaging by acquiring linear translation projection data in two vertical directions (0 degrees and 90 degrees).

[0056] (6) Analytical reconstruction algorithm is an image reconstruction method, which is different from iterative reconstruction algorithm. In the embodiments of this application, analytical reconstruction algorithm refers to an algorithm that uses projection data obtained by orthogonal line scanning and directly calculates three-dimensional image through a specific mathematical model. It has the characteristics of fast calculation speed and suitability for real-time detection in pipeline.

[0057] The embodiments of this application will be described in detail below.

[0058] CT technology is a non-destructive testing technique based on the attenuation characteristics of X-rays penetrating matter. It can reconstruct tomographic images of the interior of an object based on the attenuation information of X-rays transmitted from different directions, and is widely used in medical imaging diagnosis, industrial non-destructive testing and other fields.

[0059] With the rapid development of computer technology, microelectronics technology, and integrated circuit technology, high-density large-scale integrated circuits such as multilayer circuit boards (PCBs) and ball grid array (BGA) integrated chips are being widely used in modern electronic products. CT technology can be used to detect defects in these components, leading to the development of ODCT technology in the industry. In this technology, the X-ray source and detector move parallel to each other in a straight line, scanning the object along two mutually orthogonal linear directions. This obtains the projection data of the object in these two orthogonal directions, and image reconstruction is performed based on this data, thus achieving high-resolution imaging of the object in these two orthogonal directions.

[0060] Since ODCT technology is essentially a finite-angle CT technique, it acquires truncated projection data and reconstructs the three-dimensional structure of components using finite-angle CT reconstruction algorithms. For finite-angle reconstruction problems, the projection data often fails to meet the data integrity requirements of analytical algorithms. Therefore, iterative algorithms, which have lower data integrity requirements, are more commonly used in effective-angle CT reconstruction. Iterative algorithms require significantly more storage space and have a much longer computation time than analytical algorithms. Therefore, most ODCT imaging currently uses iterative algorithms, sacrificing storage space and runtime to achieve high-quality reconstructed images.

[0061] To address this, this application provides an ODCT image reconstruction method, apparatus, and program product. Based on ODCT, it proposes an analytical image reconstruction algorithm framework. The core of this framework is to perform weighted, filtered, back-projected, and fused scanning projection data of two mutually orthogonal straight lines through a unified projection domain coordinate mapping to perform local reconstruction of the target object. This addresses the problem of slow ODCT image reconstruction speed and enables rapid ODCT image reconstruction while ensuring a certain level of quality and stability.

[0062] It is worth noting that this application is applicable to industrial inspection scenarios, such as imaging plate-like components to facilitate defect detection based on the imaging results, but it is not limited thereto. Furthermore, this application is applicable to image reconstruction based on ODCT technology, and can also be used for image reconstruction based on ODCT derivative technologies such as OTCL technology, without specific limitations.

[0063] Reference Figure 1 The ODCT image reconstruction method provided in this application embodiment may include the following steps S101-S103:

[0064] S101, Perform orthogonal linear scanning of the target object along the horizontal and vertical axes to obtain the first projection data and the second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis.

[0065] It is understandable that the target object is the object to be reconstructed, such as PCB, BCA, etc., but is not limited to this.

[0066] Specifically, the X-ray source and the flat panel detector move relative to each other, performing orthogonal linear scanning of the target object along two mutually perpendicular straight lines (i.e., along the X-axis and Z-axis of the reconstruction plane, where the X-axis is the horizontal axis and the Z-axis is the vertical axis). The data acquisition device and the reconstruction coordinate system are set up as follows: Figure 2 As shown, each scanning trajectory covers the same angular range, acquiring a sequence of projected images from the corresponding viewpoint. The projected data obtained by scanning along the X-axis straight line trajectory is defined as the first projection data, denoted as... The projection data obtained by scanning along the Z-axis linear trajectory is defined as the second projection data, denoted as... .

[0067] It should be noted that, Figure 2 It includes a radiation source, a flat panel detector, and the object being measured. SOD is the vertical distance from the radiation source to the object, and SDD is the vertical distance from the radiation source to the detector. For example... Figure 2 As shown, the object under test is fixed between the X-ray source and the detector. The X-ray source is below the object and emits X-rays upwards. The flat panel detector is above the object and receives the projection of the X-rays after they penetrate the object. Simultaneously, it maintains relative motion with the X-ray source during its movement (i.e., the center of the X-ray source, the center of the object under test, and the center of the detector are always on a straight line). The path shown is... The diagram illustrates the linear scanning path of the X-ray source along the X-axis. The diagram illustrates the linear scanning path along the Z-axis, while the corresponding linear scanning path for the detector is... and This system enables orthogonal linear scanning of the object being measured.

[0068] Here, the embodiments of this application can collect projection data of the target object from two dimensions by performing orthogonal scanning along the horizontal and vertical axes, thereby covering the key angle range of the reconstruction plane, avoiding the information blind spots of single-direction scanning, and providing a comprehensive raw data foundation for subsequent image reconstruction.

[0069] Alternatively, in one example, the first projection data and the second projection data can be acquired through a system that:

[0070] A platform for placing and rotating target objects;

[0071] An X-ray source is used to emit X-rays toward a target object;

[0072] A flat panel detector is used to collect projection data of a target object when an X-ray source emits X-rays.

[0073] Here, the target object can be placed on a platform. First, a horizontal axis scan is performed: the X-ray source is activated, emitting X-rays towards the target object. During this process, a flat panel detector collects the projection data of the target object in the horizontal axis direction, i.e., the first projection data. Then, a vertical axis scan is performed: the platform is rotated 90 degrees, and the X-ray source is activated, emitting X-rays towards the target object. During this process, a flat panel detector collects the projection data of the target object in the vertical axis direction, i.e., the second projection data. It should be understood that the order of the horizontal and vertical axis scans can be interchanged; this embodiment does not specifically limit this. In this way, both the first and second projection data can be acquired.

[0074] Alternatively, in another example, refer to Figure 3 and Figure 4 , Figure 3 and Figure 4 The Y-axis shown is the same as the vertical Z-axis mentioned earlier. Figure 4 It includes a radiation source, a flat panel detector, and a target object. DSO is the vertical distance from the radiation source to the target object, and DSD is the vertical distance from the radiation source to the detector. Figure 4 The X-ray source is fixed below the conveyor belt and emits X-rays upwards. The flat panel detector is fixed above the target object to receive the X-ray projection after penetrating the object. Path P1 indicates the linear scanning path of the target object along the Y-axis on the first linear conveyor belt 100 equipped with the first imaging device. Path P2 indicates the linear scanning path of the target object along the X-axis on the second linear conveyor belt 400 equipped with the second imaging device. Through the continuous translational movement of the target object along paths P1 and P2, in conjunction with the fixed X-ray source and detector, the system effectively achieves ODCT scanning of the target object. The first projection data and the second projection data can be acquired by a special imaging system.

[0075] Specifically, the system may include a first linear conveyor belt 100, a right-angle steering component, and a second linear conveyor belt 400. The first linear conveyor belt 100 is disposed on the longitudinal axis, and the second linear conveyor belt 400 is disposed on the transverse axis. The discharge end of the first linear conveyor belt 100 is fixedly connected to the feed end of the second linear conveyor belt 400 through the right-angle steering component.

[0076] The first linear conveyor belt 100 is equipped with a first imaging device. The first imaging device is used to scan the target object in the longitudinal direction during the process of the target object being conveyed from the feed end of the first linear conveyor belt 100 to the discharge end of the first linear conveyor belt 100, so as to obtain the first projection data of the target object.

[0077] The second linear conveyor belt 400 is equipped with a second imaging device. The second imaging device is used to scan the target object in the horizontal direction during the process of the target object being conveyed from the feed end of the second linear conveyor belt 400 to the discharge end of the second linear conveyor belt 400, so as to obtain the second projection data of the target object.

[0078] Specifically, in this system, the first linear conveyor belt 100, the right-angle steering assembly, and the second linear conveyor belt 400 together constitute an L-shaped conveyor belt transmission mechanism. In this mechanism, the first linear conveyor belt 100 is positioned on the Y-axis, and the second linear conveyor belt 400 is positioned on the X-axis. Moreover, the discharge end of the first linear conveyor belt 100 is fixedly connected to the feed end of the second linear conveyor belt 400 through the right-angle steering assembly. This design allows the first linear conveyor belt 100 and the second linear conveyor belt 400 to be arranged at a 90-degree angle on the horizontal plane, forming an L-shaped layout.

[0079] More specifically, the first linear conveyor belt 100, serving as the system's feeding area, extends along the Y-axis and is responsible for providing scanning motion in the Y-axis direction. A right-angle steering component is located at the intersection of the first linear conveyor belt 100 and the second linear conveyor belt 400. When the target object reaches the intersection, it smoothly transitions from the Y-axis to the X-axis via the right-angle steering component. During this process, the target object's physical posture remains unchanged; that is, it does not rotate horizontally, only changing the direction of its motion vector. The second linear conveyor belt 400, serving as the system's discharging area, extends along the X-axis and forms a 90-degree angle with the first linear conveyor belt 100. Its structural configuration is the same as the first linear conveyor belt 100, and it is responsible for providing scanning motion in the X-axis direction. Optionally, both the first linear conveyor belt 100 and the second linear conveyor belt 400 are made of materials with low X-ray attenuation coefficients and have uniform belt thickness to reduce imaging background noise.

[0080] Therefore, the principle of the L-shaped conveyor belt mechanism is as follows: the target object first moves at a constant speed along the Y-axis on the first linear conveyor belt 100, then reaches the intersection of the first linear conveyor belt 100 and the second linear conveyor belt 400, i.e., the right-angle turning component, and smoothly transitions to the second linear conveyor belt 400, subsequently moving at a constant speed along the X-axis on the second linear conveyor belt 400. Throughout the entire transmission process, the system does not require the target object to be rotated in place; the target object only needs to be continuously translated along the conveyor belt, thus achieving continuous feeding in an assembly line manner.

[0081] Furthermore, the system includes a dual-station fixed imaging system along the conveyor path. This imaging system has two independent imaging areas, located on the paths of the two conveyor belts respectively, corresponding to two orthogonal scanning directions (X-axis and Y-axis). Specifically, the first linear conveyor belt 100 is equipped with a first imaging device, and the second linear conveyor belt 400 is equipped with a second imaging device. When the target object passes through the first imaging device along the Y-axis, the first imaging device operates, scanning the target object in the Y-axis direction to obtain the first projection data of the target object. When the target object turns and passes through the second imaging device along the X-axis, the second imaging device operates, scanning the target object in the X-axis direction to obtain the second projection data of the target object.

[0082] Furthermore, the first imaging device includes:

[0083] The first imaging station is used to scan the target object along the longitudinal axis during the process of the target object being conveyed from the feed end of the first linear conveyor belt 100 to the discharge end of the first linear conveyor belt 100, so as to obtain multiple first projection images of the target object.

[0084] The first encoder is used to acquire multiple longitudinal axis displacement data of the target object during the process of the target object being conveyed from the feed end of the first linear conveyor belt 100 to the discharge end of the first linear conveyor belt 100.

[0085] The first controller is used to associate each first projection image with each vertical axis displacement data to generate first projection data.

[0086] Furthermore, the first imaging station includes:

[0087] The first radiation source 200 is fixedly installed on the frame beam below the first linear conveyor belt 100. The first radiation source 200 is used to emit radiation beams, the center of which is vertically upward and has a cone-shaped geometry.

[0088] The first flat panel detector 300 is fixedly suspended above the first linear conveyor belt 100. The center of the first flat panel detector 300 is aligned with the focal point of the first radiation source 200. The first flat panel detector 300 is used to acquire multiple frames of first projection images when the first radiation source 200 emits a radiation beam.

[0089] Specifically, the first imaging device includes a first imaging station, a first encoder, and a first controller. The first imaging station is installed in the middle section of the first linear conveyor belt 100. The first imaging station includes a first X-ray source 200 and a first flat panel detector 300. The first X-ray source 200 is fixedly installed on the frame beam below the first linear conveyor belt 100 and is used to emit a X-ray beam. The center of the X-ray beam is vertically upward, forming a cone-beam geometry. The first flat panel detector 300 is a large-area digital detector, fixedly suspended above the first linear conveyor belt 100. The effective imaging area covers the full width of the first linear conveyor belt 100, and the center of the first flat panel detector 300 is strictly aligned with the focal point of the first X-ray source 200. To ensure high system stability and mechanical simplification, both the first X-ray source 200 and the first flat panel detector 300 are fixedly installed without rotating parts. Furthermore, the relative positions of the first X-ray source 200 and the first flat panel detector 300 are absolutely fixed, forming a cone-beam imaging geometry that vertically passes through the conveyor belt.

[0090] In practical applications, when the target object passes through the first imaging station along the Y-axis, the first X-ray source 200 continuously emits a X-ray beam, and the first flat panel detector 300 continuously acquires data at a high frame rate. During this process, the target object displaces relative to the first X-ray source 200, which is geometrically equivalent to the object being stationary. The first X-ray source 200 and the first flat panel detector perform a linear trajectory scan of the target object along the Y-axis. This allows for the acquisition of multiple frames of the first projection image of the target object. Furthermore, during this process, the first encoder continuously acquires the displacement data of the target object in the Y-axis direction, thereby obtaining multiple longitudinal axis displacement data of the target object.

[0091] In the above process, the first controller precisely controls the pulse emission of the X-ray source and the frame acquisition of the detector, and receives pulse signals sent by the first encoder. After obtaining multiple frames of first projected images of the target object and multiple longitudinal displacement data, the first controller associates each frame of the first projected image with each longitudinal displacement data to ensure that each projected image corresponds to a unique physical space coordinate, thereby generating first projection data. It should be understood that the first projection data includes multiple first projected images and the longitudinal displacement data corresponding to each first projected image. In this way, object scanning in the Y-axis direction can be achieved.

[0092] Furthermore, the second imaging device includes:

[0093] The second imaging station is used to scan the target object in the horizontal direction during the process of the target object being conveyed from the feed end of the second linear conveyor belt 400 to the discharge end of the second linear conveyor belt 400, so as to obtain multiple frames of second projection images of the target object.

[0094] The second encoder is used to acquire multiple transverse displacement data of the target object during the process of the target object being conveyed from the feed end of the second linear conveyor belt 400 to the discharge end of the second linear conveyor belt 400.

[0095] The second controller is used to associate each second projection image with each horizontal axis displacement data to generate second projection data.

[0096] Furthermore, the second imaging station includes:

[0097] The second radiation source 500 is fixedly installed on the frame beam below the second linear conveyor belt 400. The second radiation source 500 is used to emit radiation beams, the center of which is vertically upward and has a cone-shaped geometry.

[0098] The second flat panel detector 600 is fixedly suspended above the second linear conveyor belt 400. The center of the second flat panel detector 600 is aligned with the focal point of the second radiation source 500. The second flat panel detector 600 is used to acquire multiple frames of second projection images when the second radiation source 500 emits a radiation beam.

[0099] Specifically, the second imaging device includes a second imaging station, a second encoder, and a second controller. The second imaging station is installed in the middle section of the second linear conveyor belt 400. The second imaging station includes a second X-ray source 500 and a second flat panel detector 600. The second X-ray source 500 is fixedly installed on the frame beam below the second linear conveyor belt 400 and is used to emit a X-ray beam. The center of the X-ray beam is vertically upward, forming a cone-beam geometry. The second flat panel detector 600 is a large-area digital detector, fixedly suspended above the second linear conveyor belt 400. The effective imaging area covers the full width of the second linear conveyor belt 400, and the center of the second flat panel detector 600 is strictly aligned with the focal point of the second X-ray source 500. To ensure high system stability and mechanical simplification, both the second X-ray source 500 and the second flat panel detector 600 are fixedly installed without rotating parts. Furthermore, the relative positions of the second X-ray source 500 and the second flat panel detector 600 are absolutely fixed, forming a cone-beam imaging geometry that vertically passes through the conveyor belt.

[0100] In practical applications, when the target object passes through the second imaging station along the X-axis, the second X-ray source 500 continuously emits a beam of X-rays, and the second flat panel detector 600 continuously acquires data at a high frame rate. During this process, the target object displaces relative to the second X-ray source 500, which is geometrically equivalent to the object being stationary. The second X-ray source 500 and the second flat panel detector perform a linear trajectory scan of the target object along the X-axis. This allows for the acquisition of multiple frames of the second projection image of the target object. Furthermore, during this process, the second encoder continuously acquires the displacement data of the target object in the X-axis direction, thereby obtaining multiple transverse axis displacement data of the target object.

[0101] In the above process, the second controller precisely controls the pulse emission of the X-ray source and the frame acquisition of the detector, and receives pulse signals sent by the second encoder. After obtaining multiple frames of second projection images of the target object and multiple horizontal axis displacement data, the second controller associates each frame of second projection image with each horizontal axis displacement data to ensure that each projection image corresponds to a unique physical space coordinate, thereby generating second projection data. It should be understood that the second projection data includes multiple second projection images and the horizontal axis displacement data corresponding to each second projection image. In this way, object scanning in the X-axis direction can be achieved.

[0102] Furthermore, the aforementioned system also includes:

[0103] The first servo motor is used to control the transmission of the first linear conveyor belt 100 to transport the target object from the feed end of the first linear conveyor belt 100 to the discharge end of the first linear conveyor belt 100.

[0104] The second servo motor is used to control the transmission of the second linear conveyor belt 400 to transport the target object from the feed end of the second linear conveyor belt 400 to the discharge end of the second linear conveyor belt 400.

[0105] Specifically, in this system, when a target object enters the feed end of the first linear conveyor belt 100, the first servo motor is controlled to operate, causing the first linear conveyor belt 100 to drive and transport the target object from the feed end to the discharge end of the first linear conveyor belt 100. When the target object enters the feed end of the second linear conveyor belt 400, the second servo motor is controlled to operate, causing the second linear conveyor belt 400 to drive and transport the target object from the feed end to the discharge end of the second linear conveyor belt 400. In this way, the target object only needs to be continuously translated and transported along the conveyor belt.

[0106] Here, on the one hand, the aforementioned system utilizes an L-shaped conveyor belt as the scanning motion mechanism, directly converting the natural transport motion of the target object on the conveyor into the physical displacement required for CT scanning. During the detection process, the object does not need to stop and can flow continuously across the scanning area, one after the other. This on-the-go scanning mode greatly improves detection efficiency. On the other hand, the system innovatively adopts a fully fixed source-detector architecture, meaning that both X-ray sources and two large-area array detectors are fixedly installed without any rotating or reciprocating parts. The system relies solely on the unidirectional movement of the conveyor belt, which greatly reduces the complexity of mechanical design, lowers detection costs, and eliminates the impact of rotational vibration on accuracy, thereby improving both detection accuracy and equipment reliability and lifespan.

[0107] S102, perform weighting, filtering and back-projection operations on the first projection data and the second projection data to obtain the back-projection data of the first projection data and the back-projection data of the second projection data.

[0108] Specifically, the first projection data is weighted, filtered, and back-projected to obtain the back-projected data of the first projection data, and the second projection data is weighted, filtered, and back-projected to obtain the back-projected data of the second projection data.

[0109] Here, in this embodiment, weighted operations correct for projection geometric distortions caused by the cone angle and scanning trajectory of X-rays during scanning, and compensate for differences in energy attenuation when rays from different paths penetrate an object. Filtering operations enhance high-frequency components in the projection data and suppress stellar artifacts, thus providing high-quality data for subsequent backprojection and fusion operations. Backprojection maps the filtered projection data back to the reconstructed coordinate system, thereby initially restoring the three-dimensional spatial structure of the object and obtaining reconstructed images in two orthogonal directions. It should be understood that the backprojection data of the first projection data refers to the reconstructed image on the horizontal axis, while the backprojection data of the second projection data refers to the reconstructed image on the vertical axis.

[0110] S103, perform a fusion operation on the backprojection data of the first projection data and the backprojection data of the second projection data to obtain the reconstructed image data of the target object.

[0111] Specifically, after performing weighting, filtering, and backprojection operations, the backprojection data of the first projection data and the backprojection data of the second projection data are fused into the reconstructed image data of the target object, thereby achieving rapid ODCT image reconstruction while ensuring a certain level of quality and stability.

[0112] Here, orthogonal cross-track scanning cannot achieve full-angle scanning, and the reconstructed image in a single direction suffers from information loss, resulting in a need to improve image reconstruction quality. To address this, this application's embodiment directly adds the reconstruction results from two orthogonal directions, effectively compensating for blind spots in their respective viewpoints and improving the overall integrity of the reconstructed image. This allows for rapid ODCT image reconstruction while maintaining a certain level of quality and stability.

[0113] In some optional embodiments, step S102 above includes the following steps S201-S203, wherein the target projection data is either the first projection data or the second projection data:

[0114] S201, perform a weighted operation on the target projection data to obtain the weighted target projection data.

[0115] Furthermore, in some optional embodiments, step S201 above includes:

[0116] Calculate the weighting coefficients of the target projection data based on the scanning parameters related to orthogonal linear scanning;

[0117] The target projection data is weighted based on the weighting coefficients to obtain the weighted target projection data.

[0118] Furthermore, in some optional embodiments, the scanning parameters include the movement distance of the X-ray source, the physical distance of the detector in the first direction, the physical distance of the detector in the second direction, the distance from the X-ray source to the target object, and the distance from the plane where the X-ray source moves to the plane where the detector moves. The first direction is a direction parallel to the orthogonal linear scanning trajectory, and the second direction is a direction perpendicular to the orthogonal linear scanning trajectory.

[0119] The above calculation of weighting coefficients for target projection data based on scanning parameters related to orthogonal linear scanning includes:

[0120] The cosine value of the cone angle is calculated based on the distance from the plane of motion of the X-ray source to the plane of motion of the detector and the physical distance of the detector in the second direction.

[0121] The square root operation is performed based on the cosine value of the cone angle, the distance from the ray source to the target object, the distance from the plane where the ray source moves to the plane where the detector moves, the distance the ray source moves, and the physical distance of the detector in the first direction, to obtain the square root operation value.

[0122] The ratio of the distance from the X-ray source to the target object to the square root value is determined as the weighting coefficient.

[0123] For ease of understanding, the above weighting coefficients can be expressed as the following formula (1):

[0124] (1);

[0125] In equation (1), Indicates the weighting coefficient; Indicates the distance the radiation source has traveled; Indicates that the detector is in Physical distance in the direction (first direction); This represents the distance from the plane containing the trajectory of the ray source to the plane containing the center of the target object; This represents the distance from the plane containing the trajectory of the X-ray source to the plane containing the trajectory of the detector; The cone angle of the X-ray source during the scanning process is expressed by the following formula (2):

[0126] (2);

[0127] In equation (2), Indicates that the detector is in Physical distance in the direction (second direction).

[0128] Furthermore, in some optional embodiments, the above-mentioned weighting operation on the target projection data based on weighting coefficients to obtain weighted target projection data includes:

[0129] The weighted target projection data is determined by multiplying the weighting coefficients and the target projection data.

[0130] For ease of understanding, the weighted target projection data described above can be expressed as the following formula (3):

[0131] (3);

[0132] In equation (3), This represents the weighted target projection data; This represents the target projection data.

[0133] Here, the weighting operation is a pre-processing systematic error correction operation targeting the inherent geometric defects of the cone beam. Essentially, it normalizes and corrects the differences in projection path lengths of rays at different tilt angles, thereby compensating for projection data distortion caused by cone beam divergence and providing uniform and accurate projection data for subsequent filtering operations. Specifically, on the one hand, it effectively corrects projection geometric distortions caused by the cone angle and scanning trajectory during X-ray scanning, ensuring the spatial consistency of ray projection data at different angles; on the other hand, it compensates for the differences in energy attenuation when rays of different paths penetrate an object, making the grayscale values ​​of the target projection data more realistically reflect the density distribution inside the target object, reducing reconstruction errors caused by attenuation, and thus providing high-quality input for subsequent filtering and back-projection operations.

[0134] S202, perform filtering operation on the weighted target projection data to obtain filtered target projection data.

[0135] Furthermore, in some optional embodiments, the above-described filtering operation on the weighted target projection data to obtain filtered target projection data includes:

[0136] Construct a ramp filter for the target projection data;

[0137] Perform a fast Fourier transform on the weighted target projection data to obtain the transformed target projection data;

[0138] The transformed target projection data is processed using a ramp filter to obtain the processed target projection data.

[0139] The processed target projection data is subjected to inverse Fourier transform to obtain filtered target projection data.

[0140] For ease of understanding, the filtered target projection data can be expressed as the following formula (4):

[0141] (4);

[0142] In equation (4), The corresponding one-dimensional Fourier transform function is , The variable representing the frequency domain space indicates the ramp filter; This represents the coordinates of the reconstructed point in the reconstructed coordinate system. Represents all points that have undergone reconstruction The index of the rays on the detector; The distance factor representing the reconstructed point on the ray is described in the following embodiments; This represents the filtered target projection data.

[0143] Here, filtering is the core operation for restoring high-frequency information from the projection data to the image space. Essentially, it involves performing filtering row-by-row along the detector array. Through frequency domain high-frequency compensation, it enhances key information such as object boundaries and fine structures in the projection data, restoring high-frequency details and suppressing star-shaped artifacts during backprojection, thus laying the foundation for accurate backprojection. Specifically, the ramp filter enhances the high-frequency components in the target projection data, highlighting the edges and details of the target object, thereby solving the image blurring problem caused by direct backprojection. Combined with convolution integral operations, it can further suppress random noise introduced during scanning, thus aiding in the removal of star-shaped artifacts during backprojection. In this way, the target projection data can be converted into data more suitable for backprojection, providing high-quality input for subsequent backprojection operations.

[0144] S203, perform a back-projection operation on the filtered target projection data to obtain the back-projection data of the target projection data.

[0145] Furthermore, in some optional embodiments, the above-described back-projection operation on the filtered target projection data to obtain back-projected data of the target projection data includes:

[0146] Based on the ordinate of the reconstructed point in the reconstruction coordinate system in the target projection data, combined with the distance from the ray source to the target object and the distance from the plane where the ray source moves to the plane where the detector moves, the distance factor of the reconstructed point in the target projection data on the ray is obtained.

[0147] By using the distance factor, a back-projection operation is performed on the filtered target projection data to obtain the back-projected data of the target projection data.

[0148] Further, in some optional embodiments, the distance factor of the reconstructed point on the ray in the target projection data is obtained by combining the ordinate of the reconstructed point in the target projection data in the reconstruction coordinate system with the distance from the ray source to the target object and the distance from the plane where the ray source moves to the plane where the detector moves, including:

[0149] The distance from the ray source to the target object is obtained by summing the distance from the ray source to the target object and the ordinate of the reconstructed point in the reconstructed coordinate system in the target projection data.

[0150] The distance factor is defined as the ratio of the target distance to the distance between the plane where the X-ray source is moving and the plane where the detector is moving.

[0151] For ease of understanding, the above distance factors satisfy the following formula (5):

[0152] (5).

[0153] The above back projection data can be expressed as the following formula (6):

[0154] (6);

[0155] In equation (6), This represents back-projection data.

[0156] Here, the backprojection operation is the key operation that transforms the filtered projection data into a linear attenuation coefficient distribution in three-dimensional voxel space. Its essence is based on orthogonal linear scanning trajectories, weighted by factors... The target projection data is mapped to the corresponding 3D voxels using weights. By integrating and accumulating the projection contributions at different angles, the structural information of the object along the straight-line scanning direction is initially restored, thereby correcting the influence of the geometric distance difference between the reconstructed voxels and the ray source on the intensity of the projection contribution. Specifically, the integration operation in the backprojection process can reverse-map the target projection data to the reconstruction coordinate system, initially restoring the 3D spatial structure of the object. The weighting factors involved in the integration operation... It can correct the difference in geometric distance between the reconstructed voxel and the ray source on the intensity of the projection, ensuring that the density information of the object structure is accurately restored, thus providing high-quality input for subsequent fusion operations.

[0157] In some optional embodiments, step S103 above includes the following step S301:

[0158] S301, the sum of the backprojection data of the first projection data and the backprojection data of the second projection data is determined as the reconstructed image data.

[0159] Specifically, the backprojection data of the first projection data is represented as follows: The back projection data of the second projection data is represented as Data obtained through back projection and During the orthogonal cross trajectory scanning process, the projection angle can only cover a maximum of 180 degrees, not 360 degrees. Therefore, each ray is measured at most once along different directions, and the projection data has no redundancy. Thus, the final reconstruction result is considered to be the sum of the back projection data along two orthogonal directions, as shown in the following formula (7):

[0160] (7);

[0161] In equation (7), This represents the reconstructed image data.

[0162] Here, the essence of the fusion operation lies in using the superposition of reconstruction results from two orthogonal directions to compensate for the insufficient coverage of projection data from orthogonal linear scanning. Since ODCT uses linear trajectory scanning, the projection angle can only cover a maximum of 180 degrees. A single direction cannot meet the complete coverage of the three-dimensional Radon space. By superimposing the two reconstruction results through two orthogonal scans along the horizontal and vertical axes, the spatial coverage of the projection data can be improved, and finite angle artifacts can be suppressed, thereby improving the quality of the reconstructed image.

[0163] In some optional embodiments, the above method further includes:

[0164] Set pixel values ​​less than zero in the reconstructed image data to zero.

[0165] The following section, based on the MATLAB simulation platform, ASTRA3D cone-beam CT toolbox, and CUDA parallel acceleration, uses a three-dimensional regular spherical array virtual phantom as the test object and, in conjunction with the method proposed in the embodiments of this application, fully elaborates on the specific implementation process of the embodiments of this application.

[0166] 1. Implementation preparation:

[0167] (1.1) Algorithm parameter settings:

[0168] a. Downsampling coefficient UnSam=4: Used to balance detector resolution and computational efficiency, and reduce data volume. It can be adjusted as needed.

[0169] b. Number of projection viewpoints viewNum=256: Total number of projections for orthogonal scanning, evenly distributed across the X / Z scanning directions;

[0170] c. Scan range scan_range=50.00mm: The translation range of the X-ray source along the scanning direction;

[0171] d. Detector pixel size detPitch = 0.085 × UnSam = 0.34 mm;

[0172] e. Core geometric parameters: distance from the X-ray source to the center of the object SOD = 50 mm, distance from the X-ray source to the detector SDD = 250 mm;

[0173] f. Voxel dimension parameters nVoxel=[400;3072;2400] / UnSam=[100;768;600], dimension definition: [Y,X,Z];

[0174] g. Number of detector rows (U direction / X axis): DetX=3072 / UnSam=768;

[0175] h. Number of detector rows (V direction / Z axis): DetY=2400 / UnSam=600;

[0176] i. Effective field of view of the detector: width in the U direction 768×0.34=261.12mm, height in the V direction 600×0.34=204.00mm.

[0177] (1.2) Setting and generating virtual phantom parameters:

[0178] This embodiment uses a three-dimensional spherical array virtual phantom as the test object. The phantom is a standard, regular structure without actual physical defects, used to verify the imaging accuracy and restoration degree of the reconstruction algorithm.

[0179] a. Field of view (FOV): X-direction FOVx = DetX × detPitch / (2 × SDD / SOD), calculated to FOVx = 26.112 mm; Y-direction FOVy = FOVx × nVoxel(1) / nVoxel(2), calculated to FOVy = 3.405 mm; Z-direction FOVz = FOVx × nVoxel(3) / nVoxel(2), calculated to FOVz = 20.40 mm;

[0180] b. Voxel mesh generation: A three-dimensional voxel mesh is generated based on the ndgrid function, with the X-axis range [−26.112mm, 26.112mm], the Y-axis range [−3.405mm, 3.405mm], and the Z-axis range [−20.40mm, 20.40mm].

[0181] c. Sphere parameters: single sphere radius sphere_r = 1.5mm, sphere center spacing = 3.2mm, sphere grayscale value v = 1.0;

[0182] d. Sphere array distribution: The grid range is grid_rng=−3:1:3, generating a 7×7 regular sphere array in the XZ plane, with the Y-axis center coordinate being 0;

[0183] e. Phantom storage format: .mat file, variable name phantom_vol, data type single, dimensions 768×100×600.

[0184] (1.3) Software environment:

[0185] a. Programming language: MATLAB R2022b or later; Dependent toolboxes: ASTRA Toolbox v2.1.0 (for 3D cone-beam CT), CUDA 11.6 (GPU parallel acceleration); Core function libraries: fft / ifft (frequency domain filtering), ndgrid (voxel mesh generation), permute (dimension permutation).

[0186] b. Operating hardware: RTX 5060 Laptop GPU, 32GB RAM, supporting GPU-accelerated projection generation and back-projection reconstruction.

[0187] 2. Specific implementation:

[0188] (2.1) Data acquisition stage:

[0189] 1) Construct orthogonal scan geometry vectors:

[0190] Based on the algorithm parameter settings—total projection number viewNum=256—the first 128 views are set as X-axis linear scans, and the next 128 views are set as Z-axis linear scans. The mode of "fixed virtual phantom under test, relative synchronous movement of X-ray source and detector" is adopted. The scanning range of the linear trajectory is determined according to the parameter scan_range, and scanning begins along two orthogonal linear trajectories:

[0191] During the X-axis scanning phase, the X-ray source is translated along the X-axis, as follows:

[0192] source_positions_x=linspace(−scan_range / 2,scan_range / 2,viewNum / 2);

[0193] During the Z-axis scanning phase, the X-ray source translates along the Z-axis, with the following position:

[0194] source_positions_z=linspace(−scan_range / 2,scan_range / 2,viewNum / 2).

[0195] The detector and the X-ray source are translated synchronously in opposite directions to ensure alignment of the X-ray centerline. The translation amounts are as follows:

[0196] detector_shift_x=−source_positions_x×(SDD / SOD−1),

[0197] detector_shift_z=−source_positions_z×(SDD / SOD−1).

[0198] Then, a 256×12 vector matrix vectors is created, with each row storing the coordinates of the ray source [srcx,−SOD,srcz], the coordinates of the detector center [det_shift_x,SDD−SOD,det_shiftz], and the detector basis vectors det_u=[detPitch,0,0] and det_v=[0,0,detPitch], thus completing the 3D cone-beam geometry definition for orthogonal scanning.

[0199] 2) Virtual Phantom Creation:

[0200] A 3D voxel mesh [X,Y,Z] is generated using the ndgrid function, with the mesh range matching the aforementioned field of view (FOV). Then, a sphere generation function is defined to generate a 7×7 array of spheres in the XZ plane, which are accumulated to the initialized zero matrix to obtain the final phantom_vol. The phantom is saved as phantom.mat, completing the construction of the test object. The phantom is noise-free and defect-free, serving as a standard reference template.

[0201] 3) Generate projection data:

[0202] According to the algorithm parameter settings (Experiment Preparation - Algorithm Parameter Settings), the ASTRA function is called to create the projected geometry proj_geom=astra_create_proj_geom('cone_vec',DetY,DetX,vectors), and then the ASTRA function is called to create the volume geometry vol_geom=astra_create_vol_geom(nVoxel(1),nVoxel(2),nVoxel(3),-FOV,FOV,-FOVy,FOVy,-FOVz,FOVz). After that, the GPU accelerates the generation of the projected sine curve [~,sino_data]=astra_create_sino3d_cuda(phantom_vol,proj_geom,vol_geom); finally, the dimensions are permuted projections=permute(sino_data,[3,1,2]), and the final projection data dimension is [DetectorY,DetectorX,View], that is, [600,768,256], thus completing the projection data. Data acquisition. Among them, the first projection data... The second projection data is the projection data obtained by scanning along the straight line trajectory of the X-axis. This is the projection data obtained by scanning along a straight line trajectory along the Z-axis.

[0203] (2.2) Data weighting stage:

[0204] 1) Based on the algorithm parameter settings, set u=(−(DetX−1) / 2:(DetX−1) / 2)∗detPitch, v=(−(DetY−1) / 2:(DetY−1) / 2)∗detPitch, and generate a two-dimensional grid [V,U] through ndgrid to complete the generation of detector pixel coordinates.

[0205] 2) Traverse all 256 projection viewpoints and calculate weighting coefficients in stages: When i ≤ viewNum / 2 (X-axis scanning stage): Take the ray source translation t = source_positions_x(i), the projection coordinates parallel to the scanning trajectory e = U, and the projection coordinates perpendicular to the scanning trajectory cosine value of cone angle When i > viewNum / 2 (Z-axis scanning stage): Take the source translation t = source_positionsz(i − viewNum / 2), the projection coordinates parallel to the scanning trajectory e = V, and the projection coordinates perpendicular to the scanning trajectory cosine value of cone angle .

[0206] The weighting coefficients are calculated based on the above parameters and the above formula (1).

[0207] 3) Perform a weighting operation, as shown in formula (3) above, to obtain the weighted projection data. This completes the geometric distortion and cone angle correction.

[0208] (2.3) Data filtering stage:

[0209] This step employs a frequency domain filtering strategy using ramp filtering to compensate for the blurring effect of subsequent back projection and enhance high-frequency details. Differential filtering is also applied across different scanning stages.

[0210] 1) Filter kernel construction: Construct filter kernels in the U direction (column) and V direction (row) respectively: U direction filter kernel: Nu=2^nextpow2(ceil(DetX)∗2), generate frequency domain sequence fu=linspace(0,1,Nu / 2+1), ramp filter kernel ramp_u_half=fu, final filter kernel h_u=[ramp_u_half,ramp_u_half(end-1:-1:2)]; V direction filter kernel: Nv=2^nextpow2(ceil(DetY)∗2), generate frequency domain sequence fv=linspace(0,1,Nv / 2+1), ramp filter kernel ramp_v=fv, final filter kernel h_v=[ramp_v_half,ramp_v_half(end-1:-1:2)].

[0211] 2) Taking a Fourier Transform (FT) on both sides of the filtering formula shown in formula (4) above yields the following result. ,Right now The Fourier transform is equal to Fourier transform multiplied by ramp filter The frequency domain form (i.e., h_u, h_v).

[0212] 3) Staged filtering: Traversing 256 viewpoints, during the X-axis scanning stage (i.e., the first 128 viewpoints), a Fast Fourier Transform (FFT), filtering (i.e., multiplying by h_u), and an Inverse Fast Fourier Transform (IFFT) are performed on the U-direction of the detector to obtain the filtered first projection data. During the Z-axis scanning phase (i.e., the last 128 views), FFT, filtering (i.e., multiplying by h_v), and IFFT are performed on the detector in the V direction to obtain the filtered second projection data. .

[0213] 4) Output Results: Obtain the filtered projection data. This completes the blurring effect compensation and high-frequency detail enhancement of the projected data.

[0214] (2.4) Data back projection and fusion stage:

[0215] This step involves backprojecting and fusing the filtered two-stage data using ASTRA to reconstruct the image and correct grayscale attenuation in the image depth direction.

[0216] Because the 3D backprojection algorithm BP3D_CUDA accelerated in ASTRA allows for parallel operations on multiple sets of projection data, i.e., simultaneous backprojection... and Furthermore, the distance factor cannot be customized during backprojection. According to the definition It can be known that The size of the distance factor is related to the spatial location of the reconstructed voxels but not to the projection data itself; therefore, the distance factor can be used to... Separating the multiplication operation and applying it separately, the result in the formula is obtained by simultaneously solving the two equations above:

[0217] .

[0218] Therefore, we can first perform back projection to obtain... Then multiply by the distance factor Related items Achieve final image reconstruction:

[0219] 1) Adjust the projection data dimension projA=permute(projFilt,[2 3 1]) to adapt to the ASTRA 3D back projection data format. Then create ASTRA data objects projection data ID proj_id and reconstructed volume data ID rec_id, and configure the back projection algorithm BP3D_CUDA with the parameter cfg=astra_struct('BP3D_CUDA').

[0220] 2) Run the back projection algorithm and extract the reconstruction result rec, with data dimensions [X,Y,Z], i.e. [768,100,600], to obtain the fused initial 3D reconstructed image. .

[0221] 3) To calculate the distance factor The size of the relevant terms requires obtaining the spatial coordinates of the reconstructed voxels. For this purpose, a new 3D mesh [X,Y,Z] of the reconstructed voxels is generated (refer to the voxel mesh settings created by the virtual phantom) to match the spatial coordinate range of the reconstruction results.

[0222] 4) Calculate distance factor related terms:

[0223] ;

[0224] 5) Multiply the reconstruction result rec by the weighting factor. The relevant terms are obtained as rec = Wpost.*rec, which enables back-projection weighting of the reconstruction results. Finally, the negative value of rec is set to 0, i.e., rec (rec < 0) = 0, to remove negative grayscale noise in the reconstruction process, resulting in the final optimized reconstructed image F.

[0225] like Figure 5 As shown, Figure 5 The figures show the reconstruction results after the data back-projection and fusion stages in this application embodiment. Figures (a), (b), and (c) represent the XOZ, XOY, and YOZ cross-sections of the reconstructed model; Figures (d), (e), and (f) represent the XOZ, XOY, and YOZ cross-sections of the generated virtual model. As can be seen from the figures, in a finite-angle scenario with orthogonal linear scanning, this application can effectively restore the core geometric structure of the sphere array model under test. The reconstructed XOZ principal cross-section clearly presents the distribution pattern of the sphere array, highly matching the geometric layout of the real model. However, in the XOY and YOZ side cross-sections, the reconstruction results exhibit a certain degree of blurring and diffusion. Compared to the clear circular cross-sections corresponding to the real model, the detail resolution of the side cross-sections still has room for improvement. Overall, this application can achieve basic reconstruction of the target structure and can meet the conventional geometric distribution observation requirements.

[0226] In addition, refer to Figure 6 This application provides an ODCT image reconstruction apparatus, comprising:

[0227] The first processing module 401 is used to perform orthogonal linear scanning of the target object along the horizontal axis and the vertical axis to obtain the first projection data and the second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis.

[0228] The second processing module 402 is used to perform weighting, filtering and back-projection operations on the first projection data and the second projection data to obtain the back-projection data of the first projection data and the back-projection data of the second projection data.

[0229] The third processing module 403 is used to perform a fusion operation on the back projection data of the first projection data and the back projection data of the second projection data to obtain the reconstructed image data of the target object.

[0230] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0231] Finally, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described ODCT image reconstruction method.

[0232] The content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0233] In summary, this application possesses at least one of the following technical effects:

[0234] 1. Specifically address the core pain point of low efficiency in existing ODCT iterative algorithms:

[0235] Existing ODCT imaging technology relies on finite-angle CT reconstruction algorithms, but the truncation characteristics of projection data make analytical algorithms unsuitable, forcing the use of iterative algorithms with lower data integrity requirements. However, iterative algorithms inherently suffer from large storage space requirements and long computation times, contradicting the efficiency demands of industrial batch inspections and real-time inspection on production lines.

[0236] To address this, this application directly solves for the 3D reconstruction results by performing unified projection domain coordinate mapping, weighted correction, ramp filtering, directional backprojection, and overlay fusion on two sets of orthogonal projection data, thus avoiding the iterative approximation process of iterative algorithms. Compared to existing technologies, this application achieves an order-of-magnitude reduction in computation time and a significant decrease in storage space requirements while ensuring basic imaging quality (meeting the needs of defect identification and structural observation), significantly improving ODCT imaging efficiency and adapting to the high-efficiency detection needs of industrial scenarios.

[0237] 2. Balancing high resolution and imaging stability, it overcomes the performance shortcomings of existing scanning methods:

[0238] Existing single-line scanning ODCT equipment suffers from an imbalance between high resolution in the scanning direction and low resolution in the vertical scanning direction. While orthogonal line scanning can achieve high-resolution imaging in both directions, it is still limited by the efficiency bottleneck of iterative algorithms. At the same time, traditional analytical algorithms cannot be directly adapted to the limited-angle scanning scenario of ODCT due to their strict requirements on the integrity of projection data.

[0239] To address this contradiction, this application overcomes the issue through two core design features: First, it retains the structural advantages of orthogonal linear scanning (translation in both the X and Z axes) to ensure that the target object achieves high spatial resolution in both orthogonal directions; second, it derives and designs a reconstruction algorithm based on the orthogonal structure, enabling the reconstructed image to have both high resolution in both directions and maintain stable imaging quality.

[0240] 3. Compatible with existing hardware devices, reducing the cost of technology implementation and upgrades:

[0241] To improve the imaging efficiency of ODCT using existing technologies, it is often necessary to reconstruct the equipment structure and replace it with high-performance hardware (such as enhancing computing power modules and optimizing the transmission system), resulting in high investment costs and long upgrade cycles.

[0242] This application addresses this issue by requiring no changes to the core hardware of existing ODCT equipment; performance leaps can be achieved solely through algorithm-level upgrades. Users can directly deploy this algorithm on existing orthogonal linear scanning ODCT equipment without purchasing additional hardware or modifying existing equipment, significantly lowering the barriers to technology implementation and upgrade costs, and extending the lifespan of existing equipment.

[0243] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

[0244] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. An ODCT image reconstruction method, characterized in that, Includes the following steps: The target object is scanned along the horizontal and vertical axes using orthogonal straight lines to obtain first projection data and second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis. The first projection data and the second projection data are weighted, filtered and back-projected to obtain the back-projected data of the first projection data and the back-projected data of the second projection data. The back-projection data of the first projection data and the back-projection data of the second projection data are fused to obtain the reconstructed image data of the target object.

2. The method according to claim 1, characterized in that, The step of weighting, filtering, and backprojecting the first projection data and the second projection data to obtain the backprojected data of the first projection data and the backprojected data of the second projection data includes: The target projection data is weighted to obtain the weighted target projection data. The weighted target projection data is filtered to obtain the filtered target projection data. Perform a back-projection operation on the filtered target projection data to obtain the back-projected data of the target projection data; The target projection data is either the first projection data or the second projection data.

3. The method according to claim 2, characterized in that, The weighting operation on the target projection data to obtain the weighted target projection data includes: The weighting coefficients of the target projection data are calculated based on the scanning parameters associated with orthogonal linear scanning. The target projection data is weighted based on the weighting coefficients to obtain the weighted target projection data.

4. The method according to claim 3, characterized in that, The scanning parameters include the movement distance of the X-ray source, the physical distance of the detector in a first direction, the physical distance of the detector in a second direction, the distance from the plane where the X-ray source moves to the center plane of the target object, and the distance from the plane where the X-ray source moves to the plane where the detector moves. The first direction is a direction parallel to the orthogonal linear scanning trajectory, and the second direction is a direction perpendicular to the orthogonal linear scanning trajectory. The step of calculating the weighting coefficients of the target projection data based on the scanning parameters related to orthogonal linear scanning includes: The cosine value of the cone angle is calculated based on the distance from the plane of motion of the radiation source to the plane of motion of the detector and the physical distance of the detector in the second direction. The square root operation is performed based on the cosine value of the cone angle, the distance from the plane where the ray source is moving to the center plane of the target object, the distance from the plane where the ray source is moving to the plane where the detector is moving, the movement distance of the ray source, and the physical distance of the detector in the first direction to obtain the square root operation value; The weighting coefficient is determined by the ratio of the distance from the plane where the ray source is moving to the center plane of the target object to the square root value.

5. The method according to claim 2, characterized in that, The step of filtering the weighted target projection data to obtain filtered target projection data includes: Construct a ramp filter for the target projection data; Perform a Fast Fourier Transform on the weighted target projection data to obtain the transformed target projection data; The transformed target projection data is processed using the ramp filter to obtain the processed target projection data; The processed target projection data is subjected to inverse Fourier transform to obtain filtered target projection data.

6. The method according to claim 2, characterized in that, The step of performing a back-projection operation on the filtered target projection data to obtain the back-projected data of the target projection data includes: Based on the ordinate of the reconstructed point in the target projection data in the reconstruction coordinate system, combined with the distance from the plane where the ray source is moving to the center plane of the target object, and the distance from the plane where the ray source is moving to the plane where the detector is moving, the distance factor of the reconstructed point in the target projection data on the ray is obtained. Using the distance factor, a back-projection operation is performed on the filtered target projection data to obtain the back-projected data of the target projection data.

7. The method according to claim 6, characterized in that, The step of obtaining the distance factor of the reconstructed point on the ray in the target projection data based on the ordinate of the reconstructed point in the reconstruction coordinate system, combined with the distance from the plane where the ray source moves to the center plane of the target object, and the distance from the plane where the ray source moves to the plane where the detector moves, includes: The distance from the plane where the ray source is moving to the center plane of the target object and the sum of the ordinates of the reconstructed points in the target projection data in the reconstructed coordinate system are obtained as the target distance; The distance factor is defined as the ratio of the target distance to the distance between the plane of motion of the radiation source and the plane of motion of the detector.

8. The method according to claim 1, characterized in that, The step of fusing the back-projection data of the first projection data and the back-projection data of the second projection data to obtain the reconstructed image data of the target object includes: The sum of the backprojection data of the first projection data and the backprojection data of the second projection data is determined as the reconstructed image data.

9. An ODCT image reconstruction device, characterized in that, include: The first processing module is used to perform orthogonal linear scanning of the target object along the horizontal axis and the vertical axis to obtain the first projection data and the second projection data of the target object; wherein, the first projection data is the projection data of the target object on the horizontal axis, and the second projection data is the projection data of the target object on the vertical axis. The second processing module is used to perform weighting, filtering and back-projection operations on the first projection data and the second projection data to obtain the back-projection data of the first projection data and the back-projection data of the second projection data. The third processing module is used to perform a fusion operation on the back-projection data of the first projection data and the back-projection data of the second projection data to obtain the reconstructed image data of the target object.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements an ODCT image reconstruction method as described in any one of claims 1-8.

Images