Intelligent imaging auxiliary system of cold cathode ray machine

Abstract

The invention relates to the technical field of ray detection, and particularly discloses an intelligent imaging auxiliary system of a cold cathode ray machine, which comprises a multi-modal data acquisition unit for acquiring original imaging data and pose data, a motion compensation tomography reconstruction module, a defect quantitative analysis module and a decision support module, and comparing the defect parameter set with a defect parameter library preset by the system, and outputting the current detection image and a label document containing whether the defect exists or not and the type or residual life prediction. According to the method, equipment jitter is collected in real time by integrating the IMU, meanwhile, an innovative algorithm is matched, jitter information and original projection data are combined, motion artifacts are compensated in real time in the image reconstruction process, the definition of a detected image is improved, and image noise is reduced.

Description

Technical Field

[0001] This invention relates to the field of X-ray imaging and detection technology, and in particular to a cold cathode ray detection and imaging system, as well as an intelligent imaging system for intelligently processing and improving the quality of detection and imaging images, specifically to an intelligent imaging auxiliary system for a cold cathode ray machine. Background Technology

[0002] Portable X-ray inspection devices are now widely used in industrial non-destructive testing, public safety, medical diagnosis, and scientific research. Their core technologies include miniaturized X-ray tubes, high-sensitivity detectors, and intelligent imaging systems. In industry, these devices are used for weld inspection, casting defect detection, and internal structure analysis of electronic products. In security inspections, they can quickly identify dangerous items in luggage and packages. In the medical field, they support bedside orthopedic imaging and dental diagnosis. In scientific research, they are used for materials analysis and artifact authentication. These devices, through digital imaging technology and AI-assisted analysis, achieve real-time visual results output. They are lightweight, have low radiation doses, and are wirelessly operable, significantly improving on-site inspection efficiency, especially in scenarios where traditional fixed equipment cannot cover, such as field operations and emergency response.

[0003] Although portable X-ray inspection devices have significant advantages over fixed hot cathode X-ray inspection devices in terms of startup speed, energy consumption, and flexibility, portable devices are prone to motion artifacts in complex environments (such as unstable handheld operation or target movement); the signal-to-noise ratio is insufficient for low-dose imaging requirements; and professional personnel are required to interpret the images. Summary of the Invention

[0004] To address the problems of motion artifacts, insufficient signal-to-noise ratio for low-dose imaging, and the need for professional image interpretation in existing portable X-ray detection devices, this application provides an intelligent imaging assistance system for a cold cathode ray machine. Through improvements in basic motion and jitter information collected by the sensor module and algorithmic enhancements, the system significantly improves the quality and success rate of imaging handheld or moving targets, reducing the demands on operator stability. Simultaneously, the algorithm automatically optimizes exposure parameters based on preliminary information about the target object, aiming for the lowest possible dose while maintaining image quality. Furthermore, the algorithm performs real-time denoising and enhancement on low-dose imaging results. This reduces the radiation exposure risk for operators, improves image quality, and, especially in scenarios where high doses are not permitted, enables one-click intelligent exposure, simplifying the operation process.

[0005] To achieve the above objectives, the technical solution adopted in this application is as follows: A cold cathode ray machine intelligent imaging auxiliary system includes a multimodal data acquisition unit for acquiring raw imaging data and pose data. The multimodal data acquisition unit includes a cold cathode ray device and a sensor module, and also includes an adaptive imaging optimization module, which takes the raw imaging data acquired by the cold cathode ray device as input, extracts structural features to construct an equivalent thickness map, and then performs tube voltage and dose spatial modulation based on the equivalent thickness map. The motion-compensated tomographic reconstruction module takes the pose data collected from the sensor module as input, and outputs the reconstructed image after pose correction and motion-compensated projection processing. The defect quantification and analysis module performs defect detection on the reconstructed image output by the motion-compensated tomographic reconstruction module and outputs a set of defect parameters; and The decision support module compares the defect parameter set with the system's preset defect parameter library, outputs the current detection image, and a tag document containing the presence and type of defects or a prediction of remaining lifespan.

[0006] As one of the preferred options of the present invention, the adaptive imaging optimization module uses a projection integration method to extract structural features, integrating the projection value after the rotation angle between the X-ray source and the detector. The calculation method is as follows: in, Represents the pre-scanned raw projection data; θ The X-ray source-detector rotation angle is represented by ; u represents the detector pixel coordinates; the equivalent thickness map is constructed as follows: in, Represents the air projection value / reference calibration value when there are no objects; Represents equivalent material thickness; the tube voltage The adaptive calculation formula is as follows: in, Represents the minimum transistor voltage in the system. The maximum tube voltage represents the total voltage, all in kV; k represents the material attenuation coefficient; tanh represents the hyperbolic tangent function to ensure a smooth voltage transition; the dose space... Modulation is implemented using the following function. in, Represents the baseline exposure time. Represents the minimum exposure time. This represents the maximum exposure time, in milliseconds (ms). and These represent the low and high density thresholds, respectively, in mm.

[0007] As one of the preferred embodiments of the present invention, the motion-compensated tomographic reconstruction module includes a module for constructing ideal projection data. Ideal projection data is obtained through motion compensation projection equations. After motion correction, obtain the motion-corrected data. The reconstructed image is then output through filtering and backprojection. ; The ideal projection data The calculation method is as follows: The distribution of attenuation coefficients of the target reconstructed object. D represents the rotation matrix, and D represents the detector geometry mapping matrix. Represents the Dirac function, where: ; Pose correction matrix Represents the angular offset; , Represents the translation amount; Based on the pose correction matrix Calculate the corrected data as follows: in, Represents the convolution operator. Represents the inverse Fourier transform. f Represents the spatial frequency vector. j It represents the imaginary unit.

[0008] As one of the preferred embodiments of the present invention, the reconstructed image The calculation method for filtered back projection is as follows: in, Represents a ramp filter. , For frequency.

[0009] As one of the preferred embodiments of the present invention, the motion-compensated tomographic reconstruction module further includes an iterative reconstruction module, and the reconstruction algorithm used by the iterative reconstruction module is as follows: Where A represents the system matrix; W is the noise weight matrix. ; λ represents the noise suppression factor, var(P) represents the variance of the projected data, and λ represents the iteration step size.

[0010] As one of the preferred embodiments of the present invention, the defect parameter set is obtained by sequentially processing the reconstructed image f(x,y) through multi-scale feature extraction and defect segmentation. The multi-scale feature extraction is implemented using Gaussian convolution, as specifically shown below: in, The kernel is a Gaussian convolution. For scale parameters; The defect segmentation adopts the level set evolution equation. evaluate, in, For level set functions, To smooth the Dirac function, As the weight of the curvature term, Represents area constraint terms. , For the weight of the regional energy term, , This represents the average grayscale value of the defect / background area.

[0011] As one of the preferred embodiments of the present invention, the decision support module outputs the crack propagation rate. Perform lifespan prediction; , Where a represents the current crack size in mm; N represents the number of load cycles; and C and m represent material constants. Y represents the alternating stress amplitude, in MPa; Y represents the geometric shape factor.

[0012] As one of the preferred options of the present invention, the cold cathode ray device includes a ray machine and a detector, and the sensor module includes a low-cost IMU integrated on the detector or the ray machine.

[0013] Beneficial effects: 1. This invention integrates an IMU to collect device jitter in real time, and simultaneously matches it with an innovative algorithm. By combining jitter information and original projection data, it compensates for motion artifacts in real time during image reconstruction, thereby improving the clarity of the detected image and reducing image noise.

[0014] 2. This invention can automatically optimize parameters such as tube voltage, tube current, and exposure time based on pre-scanning of the target object or visual input information from the operator, aiming for the lowest possible dose while ensuring image quality. Simultaneously, the algorithm performs real-time denoising and enhancement on the low-dose imaging results.

[0015] 3. This invention employs an innovative algorithm to intelligently adjust the dosage, which can significantly reduce the radiation exposure risk for operators; improve image quality, especially in scenarios where high doses are not permitted; and provide one-click intelligent exposure, further simplifying the operation process.

[0016] 4. Deploy lightweight AI models on the equipment side to perform real-time analysis of generated X-ray images, automatically mark suspicious areas (such as defects in power equipment), and provide preliminary classification or diagnostic suggestions. This reduces reliance on the professional skills of on-site operators; improves detection efficiency and accuracy; and provides rapid decision support. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application 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 only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an overall flowchart of the present invention.

[0019] Figure 2 It is a flowchart of the decision output. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0022] Example 1: See the instruction manual appendix Figures 1-2As shown, an intelligent imaging auxiliary system for a cold cathode ray machine includes a multimodal data acquisition unit for acquiring raw imaging data and pose data. The multimodal data acquisition unit includes a cold cathode ray device and a sensor module. The cold cathode ray device includes a ray machine and a detector. The sensor module includes a low-cost IMU integrated on the detector or ray machine. The multimodal data acquisition unit is the most fundamental part of the imaging system. The cold cathode ray device is the most basic part for acquiring X-ray data of the target object and is the sole source of raw image data. The most important sensor module is the IMU; however, other sensors can also be integrated, such as gyroscopes and accelerometers, for real-time positioning or tracking of the X-ray machine and / or detector movement. Of course, this embodiment uses the simplest IMU. It is worth noting that different sensor types acquire different data and require different matching algorithms, but the final effect is basically the same: to compensate for the negative impact of equipment movement on imaging during X-ray imaging, collectively referred to as noise. The purpose of adding a sensor module is to capture the source of this noise and perform compensation / cancellation operations in the imaging data, thereby achieving noise reduction. This is a system setting commonly used in most existing optimization systems. Functionally, it significantly expands functionality. However, technically, simply adding a sensor module, regardless of the type of sensor, is largely the same. The key lies in whether the matching software algorithm can significantly improve at least one aspect, such as final image processing, processing efficiency, or dose optimization. In the system provided in this embodiment, It also includes an adaptive imaging optimization module, which takes the raw imaging data collected by the cold cathode ray device as input, extracts structural features to construct an equivalent thickness map, and then performs tube voltage and dose spatial modulation based on the equivalent thickness map. The adaptive imaging optimization module is one of the main technical innovations in this embodiment. It extracts structural features from the raw imaging data collected at low doses or from the dimensions of the object being measured as visually input by the operator, and constructs an equivalent thickness map of the components. This establishes a basis for calculating appropriate tube voltage, tube current, and exposure time based on the equivalent thickness map, thereby setting the minimum dose that meets the irradiation requirements based on the current equivalent thickness map, thus reducing actual radiation safety hazards.

[0023] The motion-compensated tomographic reconstruction module takes pose data acquired from the sensor module as input, and outputs a reconstructed image after pose correction and motion-compensated projection processing. Alternatively, it takes pose data acquired from the sensor module as input, and uses a reconstruction model, motion compensation, modified projection equations, and iterative reconstruction steps to compensate and revise the original data, obtaining reconstructed image data, thereby achieving noise reduction. For those skilled in the art, the motion compensation method and specific algorithm can be implemented using existing technologies. The preferred approach is to use algorithms and hardware configurations with low hardware requirements and low overhead that can meet the actual noise reduction objectives, facilitating flexible, low-cost edge deployment, lightweight computation, and balancing actual image clarity with ease and efficiency.

[0024] The defect quantification and analysis module performs defect detection on the reconstructed image output by the motion-compensated tomographic reconstruction module, outputting a defect parameter set. Defect quantification analysis can employ existing image recognition and segmentation methods, based on pixel value gradient distribution or normalized grayscale value differences, using boundary extraction algorithms alone or in combination to achieve defect identification. This is then compared with preset system rules or stored defect samples. Once a preset similarity threshold is reached, the corresponding defect type is output. The defect parameter set includes type, size, orientation, and corresponding risk level. The risk level is user-defined; when a preset risk level condition is met, a corresponding risk alarm is issued. The decision support module compares the defect parameter set with the system's preset defect parameter library, outputting the current detection image and a tag document containing the presence and type of defects or a prediction of remaining lifespan. Based on the aforementioned complex calculations, the decision support module performs a qualitative judgment on the currently detected object, determining whether a defect exists, and if so, how to provide feedback on the type, size, direction, and corresponding risk level of the defect. This allows operators to obtain the defect status of the currently detected object with a single click, minimizing the possibility of significant deviations in detection conclusions due to differences in the inspector's experience.

[0025] Example 2: This embodiment optimizes the algorithm in terms of dose optimization based on Embodiment 1. In this embodiment, the adaptive imaging optimization module uses a projection integration method to extract structural features, integrating the projection value after the rotation angle between the X-ray source and the detector. The calculation method is as follows: in, Represents the pre-scanned raw projection data; θ The X-ray source-detector rotation angle is represented by ; u represents the detector pixel coordinates; the equivalent thickness map is constructed as follows: in, Represents the air projection value / reference calibration value when there are no objects; Represents equivalent material thickness; the tube voltage The adaptive calculation formula is as follows: in, Represents the minimum transistor voltage in the system. The maximum tube voltage represents the total voltage, all in kV; k represents the material attenuation coefficient; tanh represents the hyperbolic tangent function to ensure a smooth voltage transition; the dose space... Modulation is implemented using the following function. in, Represents the baseline exposure time. Represents the minimum exposure time. This represents the maximum exposure time, in milliseconds (ms). and These represent the low and high density thresholds, respectively, in mm.

[0026] This embodiment extracts features from the original projection data or the data used for manual input to establish an equivalent thickness map. Then, it adjusts the corresponding tube voltage and exposure time according to the equivalent thickness map, thereby achieving the technical effects of adaptive adjustment and low-dose exposure. This minimizes unnecessary X-ray radiation damage without affecting the clarity of X-ray detection, and further reduces the energy consumption of portable X-ray devices.

[0027] Example 3: This embodiment is a further algorithm optimization based on any of the above embodiments. In this embodiment, the motion-compensated tomographic reconstruction module includes a module for constructing ideal projection data. Ideal projection data is obtained through motion compensation projection equations. After motion correction, obtain the motion-corrected data. The reconstructed image is then output through filtering and backprojection. ; The ideal projection data The calculation method is as follows: The distribution of attenuation coefficients of the target reconstructed object. D represents the rotation matrix, and D represents the detector geometry mapping matrix. Represents the Dirac function, where: ; Pose correction matrix Represents the angular offset; , Represents the translation amount; This represents the change in Z-axis rotation relative to the initial time. This represents the change in X-axis displacement relative to the initial time. This represents the change in Y-axis displacement relative to the initial moment. It's worth noting that the input data for the motion-compensated tomographic reconstruction module comes from pose data collected by the sensor module. This pose data includes the rotation angle (pitch angle) around the X-axis, denoted as... The rotation angle (yaw angle) about the Y-axis is denoted as The rotation angle (roll angle) about the Z-axis is denoted as... ,in, , , It can be acquired through the integration of gyroscope angular velocity; Represents translational displacement along the X-axis. This represents the translational displacement along the Y-axis, in mm. It provides objective evidence for the algorithm to compensate for artifacts caused by vibration and shaking, facilitating image optimization. It is important to note here that... , Although not directly in the pose correction matrix It is used as an input parameter, but it is used for 3D pose correction of projected geometry models, for example, for use in scenarios with non-coplanar scanning.

[0028] Based on the pose correction matrix Calculate the corrected data as follows: in, Represents the convolution operator. Represents the inverse Fourier transform. f Represents the spatial frequency vector. j Represents the imaginary unit. To provide a more intuitive understanding of the meaning of this invention when data is in different states, the following section discusses pre-scanned original projection data. The original projection, including hand tremors; ideal projection data. Corrected data The relationship between the three is explained in detail below: , and These are projection data in three different states, which are related to the reconstructed image. The relationships between these elements constitute the core mathematical framework of CT imaging. The following is a detailed explanation of each element: in, It is the intensity of incident X-rays when there is no object. Represents detector pixels u At angle The received intensity directly reflects the integral value of the attenuation after the radiation penetrates an object. ∈[0,2π], u∈[1,N] pixel ].

[0029] And ideal projection data With reconstructed images The relationship can be represented as follows: The function is the Dirac delta function, i.e., it is chosen to satisfy the equation of the straight line. The point where = 0 can be geometrically understood as defining a straight line with a perpendicular distance u from the detector and a normal angle θ; of course, for simplification, it can also be expressed as Let ds represent the straight-line path of the X-ray from the source to the detector pixel ds, and ds represent the differential arc length along the ray. The Radon transform is denoted as... Then there is in, Represents the motion correction operator. This represents the pose data matrix from the sensor module, as mentioned above. , , , and A general term.

[0030] Example 4: This embodiment is a further algorithm optimization based on any of the above embodiments. In this embodiment, the reconstructed image... The calculation method for filtered back projection is as follows: in, Represents a ramp filter. , For frequency.

[0031] In this embodiment, the motion-compensated tomographic reconstruction module further includes an iterative reconstruction module, and the reconstruction algorithm used by the iterative reconstruction module is as follows: Where A represents the system matrix; W is the noise weight matrix. ; λ represents the noise suppression factor, var(P) represents the variance of the projected data, and λ represents the iteration step size.

[0032] In this embodiment, the defect parameter set is obtained by sequentially processing the reconstructed image f(x,y) through multi-scale feature extraction and defect segmentation. The multi-scale feature extraction is implemented using Gaussian convolution, as specifically shown below: in, The kernel is a Gaussian convolution. For scale parameters; The defect segmentation adopts the level set evolution equation. evaluate, in, For level set functions, To smooth the Dirac function, As the weight of the curvature term, Represents area constraint terms. , For the weight of the regional energy term, , This represents the average grayscale value of the defect / background area.

[0033] In this embodiment, the decision support module outputs the crack propagation rate. Perform lifespan prediction; , Where a represents the current crack size in mm; N represents the number of load cycles; and C and m represent material constants. Y represents the alternating stress amplitude, in MPa; Y represents the geometric shape factor.

[0034] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A cold cathode ray machine intelligent imaging auxiliary system, comprising a multimodal data acquisition unit for acquiring raw imaging data and pose data, wherein the multimodal data acquisition unit includes a cold cathode ray device and a sensor module, characterized in that: Also includes The adaptive imaging optimization module takes the raw imaging data acquired by the cold cathode ray device as input, extracts structural features to construct an equivalent thickness map, and then performs tube voltage and dose spatial modulation based on the equivalent thickness map. The motion-compensated tomographic reconstruction module takes the pose data collected from the sensor module as input, and outputs the reconstructed image after pose correction and motion-compensated projection processing. The defect quantification and analysis module performs defect detection on the reconstructed image output by the motion-compensated tomographic reconstruction module and outputs a set of defect parameters. as well as The decision support module compares the defect parameter set with the system's preset defect parameter library, outputs the current detection image, and a tag document containing the presence and type of defects or a prediction of remaining lifespan.

2. The intelligent imaging auxiliary system for a cold cathode ray machine according to claim 1, characterized in that: The adaptive imaging optimization module uses a projection integration method to extract structural features, integrating the projection value obtained by the rotation angle between the X-ray source and the detector. The calculation method is as follows: in, Represents the pre-scanned raw projection data; θ The X-ray source-detector rotation angle is represented by ; u represents the detector pixel coordinates; the equivalent thickness map is constructed as follows: in, Represents the air projection value / reference calibration value when there are no objects; Represents equivalent material thickness; the tube voltage The adaptive calculation formula is as follows: in, Represents the minimum transistor voltage in the system. The maximum tube voltage represents the total voltage, all in kV; k represents the material attenuation coefficient; tanh represents the hyperbolic tangent function to ensure a smooth voltage transition; the dose space... Modulation is implemented using the following function. in, Represents the baseline exposure time. Represents the minimum exposure time. This represents the maximum exposure time, in milliseconds (ms). and These represent the low and high density thresholds, respectively, in mm.

3. The intelligent imaging auxiliary system for a cold cathode ray machine according to claim 2, characterized in that: The motion-compensated tomography reconstruction module includes features for constructing ideal projection data. Ideal projection data is obtained through motion compensation projection equations. After motion correction, obtain the motion-corrected data. The reconstructed image is then output through filtering and backprojection. ; The ideal projection data The calculation method is as follows: The distribution of attenuation coefficients of the target reconstructed object. D represents the rotation matrix, and D represents the detector geometry mapping matrix. Represents the Dirac function, where: ; Pose correction matrix Represents the angular offset; , Represents the translation amount; Based on the pose correction matrix Calculate the corrected data as follows: in, Represents the convolution operator. Represents the inverse Fourier transform. f Represents the spatial frequency vector. j It represents the imaginary unit.

4. The intelligent imaging auxiliary system for a cold cathode ray machine according to claim 3, characterized in that: The reconstructed image The calculation method for filtered back projection is as follows: in, Represents a ramp filter. , For frequency.

5. The intelligent imaging auxiliary system for a cold cathode ray machine according to any one of claims 3-4, characterized in that: The motion-compensated tomographic reconstruction module also includes an iterative reconstruction module, and the reconstruction algorithm used by the iterative reconstruction module is as follows: Where A represents the system matrix; W is the noise weight matrix. ; λ represents the noise suppression factor, var(P) represents the variance of the projected data, and λ represents the iteration step size.

6. The intelligent imaging auxiliary system for a cold cathode ray machine according to any one of claims 3-4, characterized in that: The defect parameter set is obtained by sequentially processing the reconstructed image f(x,y) through multi-scale feature extraction and defect segmentation. The multi-scale feature extraction is implemented using Gaussian convolution, as specifically shown below: in, The kernel is a Gaussian convolution. For scale parameters; The defect segmentation adopts the level set evolution equation. evaluate, in, For level set functions, To smooth the Dirac function, As the weight of the curvature term, Represents area constraint terms. , For the weight of the regional energy term, , This represents the average grayscale value of the defect / background area.

7. The intelligent imaging auxiliary system for a cold cathode ray machine according to any one of claims 3-4, characterized in that: The decision support module outputs the crack propagation rate. Perform lifespan prediction; , Where a represents the current crack size in mm; N represents the number of load cycles; and C and m represent material constants. Y represents the alternating stress amplitude, in MPa; Y represents the geometric shape factor.

8. The intelligent imaging auxiliary system for a cold cathode ray machine according to claim 1, characterized in that: The cold cathode ray device includes a ray machine and a detector, and the sensor module includes a low-cost IMU integrated on the detector or ray machine.

Images