A bendable DR detector

Abstract

This invention discloses a flexible DR detector, relating to the field of industrial non-destructive testing technology. It includes a DR detector body, on which an edge frame, a flexible detection plate, and stabilizing wheels are fixedly connected. A geometric distortion correction module is configured to calculate a comprehensive distortion index and trigger regional correction by quantifying the coupling effect of macroscopic bending differences and microscopic pixel deformation. A double-sided exposure compensation module is configured to quantify the non-uniformity of double-sided exposure of the flexible detection plate. A scene complexity adaptation module is configured to quantify scene complexity and adaptively switch detection modes. A control module controls the DR detector body to perform curvature parameter correction and detection mode switching. This invention achieves breakthroughs in correction accuracy, exposure uniformity, efficiency, and accuracy balance in DR detectors under dynamic bending scenarios through macroscopic + microscopic coupling to quantify geometric distortion, multi-physical parameter coupling to quantify exposure differences, and multi-dimensional coupling to quantify scene complexity.

Description

Technical Field

[0001] This invention relates to the field of industrial non-destructive testing technology, specifically a flexible DR detector. Background Technology

[0002] In the field of industrial non-destructive testing, the detection of internal defects (such as cracks and pores) in curved workpieces (such as pipes, pressure vessels, and aero-engine blades) relies on DR (digital radiography) technology. Traditional flat-panel DR detectors are rigid structures and are only suitable for detecting flat or small-curvature workpieces. However, existing flexible DR detectors achieve surface bonding through flexible substrates, thus becoming the core equipment for curved surface detection.

[0003] However, in dynamic bending scenarios, existing flexible DR detectors suffer from geometric distortions such as pixel stretching / compression due to curvature changes. Furthermore, the uneven grayscale caused by the coupling between the bending direction and the distance to the X-ray source can lead to differences in double-sided exposure. In addition, existing technologies often use a manual selection of "high-speed mode" or "high-precision mode" to switch modes, thus failing to quantify the scene complexity through multi-dimensional parameters such as curvature dispersion, material attenuation, and acquisition time. To solve the above problems, it is necessary to construct a mathematical model by quantifying parameters. Summary of the Invention

[0004] The purpose of this invention is to provide a flexible DR detector that solves the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution, including a DR detector body, on which an edge frame, a flexible detection plate, and stabilizing wheels are fixedly connected respectively, and the processor inside the DR detector body is fixedly connected to a geometric distortion correction module, a double-sided exposure compensation module, a scene complexity adaptation module, and a control module.

[0006] The geometric distortion correction module is configured to collect curvature and pixel stretching data when the flexible detection plate is bent, and calculate and generate a comprehensive distortion index by quantifying the coupling effect of macroscopic bending differences and microscopic pixel deformation, and trigger regional correction.

[0007] The double-sided exposure compensation module is configured to receive the comprehensive distortion index generated by the geometric distortion correction module, and couple the bending direction, exposure surface posture and X-ray source distance parameters of the flexible detection plate during detection to quantify the non-uniformity of the double-sided exposure of the flexible detection plate and calculate and generate the double-sided exposure direction factor.

[0008] The scene complexity adaptation module is configured to receive the double-sided exposure direction factor generated by the double-sided exposure compensation module, integrate the curvature dispersion and acquisition time dimension data during the detection of the flexible detection plate, and combine the material attenuation of the detected material to quantify the scene complexity, calculate and generate the bending complexity index, and adaptively switch the detection mode according to the bending complexity index.

[0009] The control module is configured to receive parameters calculated and generated by the geometric distortion correction module, the double-sided exposure compensation module and the scene complexity adaptation module, as well as the switched detection modes, and control the DR detector body to perform curvature parameter correction and detection mode switching.

[0010] Optionally, edge frames are fixedly connected to both sides of the flexible detection plate, a flexible handle is fixedly connected to one side of the edge frame, and a strapping hole frame is fixedly connected to the other two sides of the edge frame away from the flexible handle. The stabilizing wheel is fixedly connected to the inside of the strapping hole frame by a strapping strap fixing assembly.

[0011] Optionally, the geometric distortion correction module specifically includes:

[0012] The macro curvature difference calculation module is configured to acquire the maximum curvature and minimum curvature of the flexible probe plate when it is bent by an array of strain sensors distributed on the surface of the flexible probe plate substrate, calculate the first difference between the maximum curvature and the minimum curvature, and divide the first difference by a preset reference curvature to obtain the normalized macro curvature difference parameter.

[0013] The micro-pixel stretching calculation module is configured to measure the axial length change of the DR detector body through a grating ruler fixedly connected to the DR detector body, multiply the axial length change by a preset original pixel pitch, and then divide by the square of the reference length of the DR detector body in a squared state to obtain the micro-pixel stretching parameters.

[0014] The comprehensive distortion index generation module is configured to add the macro curvature difference parameter and the micro pixel stretching parameter to generate the comprehensive distortion index, and to divide the high-precision correction area and the high-speed correction area according to the comprehensive distortion index.

[0015] Optionally, the double-sided exposure compensation module specifically includes:

[0016] The bending direction and exposure surface coupling module is configured to determine the bending direction of the flexible detection plate during detection by a Hall element, and assign a preset bending direction coefficient according to the bending direction; determine the exposure surface of the flexible detection plate during detection by an attitude sensor, and assign a preset exposure surface efficiency coefficient to the exposure surface; and multiply the two coefficients by the exposure surface efficiency coefficient to obtain the basic exposure difference direction parameter.

[0017] The X-ray source distance correction module is configured to acquire the actual X-ray source distance through a laser rangefinder, calculate the first ratio between the actual X-ray source distance and the preset reference X-ray source distance, divide the comprehensive distortion index by the normalized maximum curvature to obtain the first weight of the curvature difference on the exposure, and multiply the first ratio by the first weight to obtain the correction parameter.

[0018] The exposure difference coefficient generation module is configured to multiply the basic exposure difference direction parameter by the correction parameter to generate the double-sided exposure direction factor.

[0019] Optionally, the scene complexity adaptation module specifically includes:

[0020] The curvature dispersion calculation module is configured to collect curvature distribution data through a strain sensor array, calculate the standard deviation of the curvature distribution, and divide the standard deviation by a preset reference curvature dispersion to obtain the normalized curvature dispersion parameter.

[0021] The exposure difference integration module is configured to obtain the absolute value of the double-sided exposure direction factor, and multiply it by the curvature dispersion normalization value and the preset material attenuation coefficient to generate the basic parameters of scene complexity.

[0022] The acquisition time weighting module is configured to acquire the actual image acquisition time through a timer, calculate the second ratio of the actual image acquisition time to the preset reference acquisition time, and multiply it by a preset weight to obtain the time efficiency parameter.

[0023] The complexity index generation module is configured to add the scene complexity base parameters output by the exposure difference integration module and the acquisition time weighting module to the time efficiency parameter to generate the bending complexity index.

[0024] The mode switching control module is configured to trigger adaptive selection of the detection mode based on the scene complexity index.

[0025] Optionally, the adaptive selection of the detection mode includes:

[0026] If the bending complexity index is less than 5, then the high-speed mode is selected;

[0027] If 5 ≤ the bending complexity index < 8, then the hybrid mode is selected;

[0028] If the bending complexity index is ≥8, then the high-precision mode is selected.

[0029] Optionally, the control module performs curvature parameter correction, including correction of the maximum and minimum curvature, with the specific correction formula as follows:

[0030] Q corrmax =QC max ×(1+WF×0.02);

[0031] Q corr min =QC min ×(1-WF×0.02);

[0032] In the formula: Q corr max Q is the adjusted maximum curvature. corr min To achieve the minimum curvature after adjustment, QC max For maximum curvature, QC min WF is the minimum curvature, and WF is the bending complexity index.

[0033] Optionally, the bending direction coefficient is set according to the outward and inward bending directions, and the bending direction coefficient is -1 when bending outward and 1 when bending inward.

[0034] The exposure surface coefficient is set according to the front and back of the exposure surface. When it is the front, the exposure surface coefficient is set to 1, and when it is the back, the exposure surface coefficient is set to 0.9.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] I. The geometric distortion correction module of this invention reflects the unevenness of curvature distribution in different regions when the detector is bent by quantifying macroscopic bending differences. It further reflects the physical spacing deviation of pixels caused by bending stretching / compression by quantifying microscopic pixel deformation. This forms a superposition of macroscopic bending differences and microscopic pixel deformation under coupled influence, resulting in a comprehensive distortion index. This achieves simultaneous coverage of macroscopic bending unevenness and microscopic pixel deformation, avoiding the phenomenon of "missed local distortion" caused by using only a single curvature value in the traditional method, and realizing accurate regional correction.

[0037] II. The dual-sided exposure compensation module of the present invention can quantify the degree of exposure unevenness by coupling physical parameters such as "bending direction, exposure surface posture, comprehensive distortion index, and X-ray source distance parameter" to achieve dynamic adaptive adjustment of exposure parameters. The combination of bending direction and exposure surface posture can directly reflect the direction of exposure unevenness. The first weight based on the comprehensive distortion index can further reflect the contribution of bending unevenness to the exposure distribution. In addition, the first ratio based on the X-ray source distance parameter can correct the exposure intensity fluctuation caused by the change of X-ray source distance. In this way, not only is the degree of exposure unevenness quantified, but the direction of overexposure / underexposure is also clearly distinguished to achieve dynamic exposure compensation.

[0038] Third, the scene complexity adaptation module of this invention achieves adaptive mode switching by quantifying scene complexity in multiple dimensions, avoiding the extreme cases of "sacrificing efficiency for accuracy" or "sacrificing accuracy for efficiency", which greatly improves the overall performance of the bendable DR detector in dynamic bending scenarios and meets the core requirements of "fast and accurate" industrial inspection. Attached Figure Description

[0039] Figure 1 This is a front view of the overall structure of the flexible DR detector.

[0040] Figure 2 For the present invention Figure 1 Enlarged schematic diagram of the structure at point A;

[0041] Figure 3 This is a schematic diagram illustrating the process of the processor inside the DR detector of the present invention performing detection.

[0042] In the diagram: 1-DR detector body, 2-edge frame, 3-flexible detector plate, 4-flexible handle, 5-binding hole frame, 6-stabilizing wheel. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1:

[0045] Please see Figures 1 to 3 This embodiment provides a flexible DR detector, including a DR detector body 1. An edge frame 2, a flexible detection plate 3, and a stabilizing wheel 6 are fixedly connected to the DR detector body 1. The edge frame 2 is fixedly connected to both sides of the flexible detection plate 3. A flexible handle 4 is fixedly connected to one side of the edge frame 2. Multiple binding holes are opened on the other two sides of the edge frame 2 away from the flexible handle 4, forming a binding hole frame 5. The stabilizing wheel 6 is fixedly connected to the inside of the binding hole frame 5 by nylon binding straps. The DR detector body 1 integrates a processor. The processor is fixedly connected to a geometric distortion correction module, a double-sided exposure compensation module, a scene complexity adaptation module, and a control module through a circuit board. The modules communicate with each other through a data bus to realize real-time data transmission and processing.

[0046] The geometric distortion correction module is configured to collect curvature and pixel stretching data when the flexible probe plate 3 is bent, and calculate and generate a comprehensive distortion index by quantifying the coupling effect of macroscopic bending differences and microscopic pixel deformation, and trigger regional correction.

[0047] The double-sided exposure compensation module is configured to receive the comprehensive distortion index generated by the geometric distortion correction module, and couple the bending direction, exposure surface attitude and X-ray source distance parameters of the flexible detection plate 3 during detection, quantify the non-uniformity of the double-sided exposure of the flexible detection plate 3, and calculate and generate the double-sided exposure direction factor.

[0048] The scene complexity adaptation module is configured to receive the double-sided exposure direction factor generated by the double-sided exposure compensation module, integrate the curvature dispersion and acquisition time dimension data during the detection of the flexible detection board 3, and combine the material attenuation of the detected material to quantify the scene complexity, calculate and generate the bending complexity index, and adaptively switch the detection mode according to the bending complexity index.

[0049] The control module is configured to receive parameters calculated and generated by the geometric distortion correction module, the double-sided exposure compensation module, and the scene complexity adaptation module, as well as the switching detection modes, and control the DR detector body 1 to perform curvature parameter correction and detection mode switching.

[0050] In this embodiment, the DR detector body 1 is a flexible detector designed specifically for industrial applications. This detector allows exposure from both sides of the flexible detection plate 3 and has a high resolution of 99 micrometers. When bent, it can conform to non-planar structures such as pipes, effectively improving image quality and reducing distortion, thereby increasing detection efficiency and imaging consistency in pipes and confined spaces. When straightened, it can also be used for X-ray inspection of flat workpieces. Regardless of whether the detection state is bent or straightened, the flexible handle 4 is used to first attach one side of the edge frame 2 to one side of the surface of the object being detected via the stabilizing wheel 6. Then, the flexible handle 4 on the other side is pulled, and the flexible detection plate 3 and the edge frame 2 on the other side are attached to the other side of the surface of the object being detected via the stabilizing wheel 6, thus completing the detection preparation before the DR detector body 1 is detected.

[0051] The geometric distortion correction module collects curvature data in real time through a strain sensor array (using MEMS strain gauges, attached to the plate surface in a matrix) mounted on the substrate surface of the flexible detector plate 3. Simultaneously, it measures the axial length change using a grating ruler fixed to the side of the DR detector body 1. This module fuses the collected curvature data with pixel stretching data to calculate the comprehensive distortion index and delineate the correction area accordingly. The dual-sided exposure compensation module receives the comprehensive distortion index and, in conjunction with a Hall element installed inside the edge frame 2, detects the magnetic field direction to determine the bending direction. The bending direction is also determined by an attitude sensor attached to the center of the back of the flexible detector plate 3. The bending direction and exposure surface attitude data are used to generate a double-sided exposure direction factor by collecting the actual distance to the X-ray source from a laser rangefinder installed at the center of the front of the DR detector body 1 and facing the X-ray source. The scene complexity adaptation module further integrates the curvature dispersion provided by the strain sensor array, the acquisition time collected by the timer integrated into the main control chip, and the material attenuation coefficient preset in the system to generate a bending complexity index, and adaptively switches the detection mode accordingly. The control module receives the output parameters of the aforementioned modules, adjusts the curvature parameters through a PID control algorithm, and sends a mode switching command to the DR detector body 1 for execution.

[0052] In summary, this embodiment achieves geometric distortion correction, exposure compensation, and adaptive mode switching in dynamic bending scenarios through the collaborative work of multiple modules, thereby improving the accuracy and efficiency of DR detectors in the inspection of curved workpieces.

[0053] Example 2:

[0054] Please see Figure 1 and Figure 3 Based on Embodiment 1, the geometric distortion correction module provided in this embodiment of the invention specifically includes:

[0055] The macro curvature difference calculation module is configured to collect the maximum curvature and minimum curvature of the flexible probe plate 3 when it is bent by the strain sensor array distributed on the substrate surface of the flexible probe plate 3, calculate the first difference between the maximum curvature and the minimum curvature, and divide the first difference by the preset reference curvature to obtain the normalized macro curvature difference parameter.

[0056] The micro-pixel stretching calculation module is configured to measure the axial length change of the DR detector body 1 by means of a grating ruler fixedly connected to the DR detector body 1, and multiply the axial length change by the preset original pixel pitch, and then divide by the square of the reference length of the DR detector body 1 in the square state to obtain the micro-pixel stretching parameters.

[0057] The comprehensive distortion index generation module is configured to add the macro curvature difference parameter and the micro pixel stretching parameter to generate a comprehensive distortion index, and divide the high-precision correction area and the high-speed correction area according to the comprehensive distortion index.

[0058] In the above embodiments, one of the core challenges of the bendable DR detector is image distortion (such as stretching, overlapping, and edge blurring) under bending conditions. The essence of distortion is the dynamic change of the detector's geometry. Specifically, the DR detector body 1 is not a fixed bending diameter in actual detection, but a continuously changing arbitrary curvature. Fixed parameter correction cannot cover all distortions. Bending not only causes overall curvature differences, but also causes small changes in pixel spacing due to stretching / compression. The two of these factors together determine the geometric distortion of the final image. The geometric distortion correction module directly quantifies the degree of this geometric distortion by calculating a comprehensive distortion index that reflects the real-time curvature difference.

[0059] In the geometric distortion correction module, the formula for calculating the overall distortion index is as follows:

[0060] ;

[0061] In the formula: Q diff To achieve a comprehensive distortion index, QC max QC represents the maximum curvature in the real-time curvature distribution matrix. min QC represents the minimum curvature in the real-time curvature distribution matrix. ref As the reference curvature, XJ0 is the original pixel pitch, and ΔL is the change in axial length of the DR detector body 1 when bent. ref This is the reference length for the DR detector body 1 when it is laid flat. This fundamental term directly reflects the "non-uniformity" of bending. This compensation term can correct pixel misalignment errors caused by length changes (such as edge pixels being stretched when bending, requiring algorithmic reduction to restore the true size), and comprehensively affects the distortion index Q. diff The larger the value, the more severe the geometric distortion caused by bending. The algorithm needs to call a stronger distortion correction model to provide a quantitative basis for subsequent image stretching / compression correction.

[0062] The division logic between the high-precision correction zone and the high-speed correction zone is based on the comprehensive distortion index Q. diff The degree of coupling between macroscopic curvature unevenness and microscopic pixel stretching, combined with the priority of detection requirements, for example: in the comprehensive distortion index Q diff When the value is greater than 0.3, it is classified as a high-precision correction zone, indicating that the macroscopic curvature difference is extremely large when the DR detector body 1 is bent, and the microscopic pixel stretching is obvious. The coupling between the two leads to severe geometric distortion, which directly affects the defect positioning accuracy. It is necessary to use a non-rigid correction algorithm to compensate for the distortion caused by small curvature fluctuations by performing nonlinear mapping on the coordinates of each pixel. The comprehensive distortion index Q diffWhen the value is between 0.1 and 0.3, it is classified as a transitional correction zone, indicating a moderate macroscopic curvature difference, slight microscopic pixel stretching, and limited but not negligible impact of distortion on accuracy. Semi-rigid correction can be enabled in this zone. Nonlinear correction should be used for localized areas with significant curvature changes (such as near pipe welds), while linear correction should be used for other areas, balancing accuracy and computational power. The overall distortion index Q... diff When the number is less than 0.1, it is divided into a high-speed correction zone, indicating that the macroscopic curvature distribution is uniform, the microscopic pixel stretching is negligible, and the distortion has little impact on the detection accuracy. Only linear stretching correction is enabled, and the overall stretching deformation is compensated by simple coordinate scaling. Among them, non-rigid correction algorithm, semi-rigid correction and linear stretching correction are all common knowledge in existing technology, and will not be elaborated on here.

[0063] In summary, this embodiment comprehensively evaluates the degree of distortion by coupling macroscopic curvature differences with microscopic pixel stretching, avoiding the omission of local distortion caused by traditional single curvature evaluation, and improving the comprehensiveness and accuracy of correction.

[0064] Example 3:

[0065] Please see Figure 1 and Figure 3 Based on Embodiment 2, the double-sided exposure compensation module provided in this embodiment of the invention specifically includes:

[0066] The bending direction and exposure surface coupling module is configured to determine the bending direction of the flexible detection plate 3 during detection by a Hall element, and assign a preset bending direction coefficient according to the bending direction. It also determines the exposure surface of the flexible detection plate 3 during detection by an attitude sensor, and assigns a preset exposure surface efficiency coefficient to the exposure surface. The two coefficients are multiplied by the exposure surface efficiency coefficient to obtain the basic exposure difference direction parameter.

[0067] The X-ray source distance correction module is configured to acquire the actual X-ray source distance through a laser rangefinder, calculate the first ratio between the actual X-ray source distance and the preset reference X-ray source distance, divide the comprehensive distortion index by the normalized maximum curvature to obtain the first weight of the curvature difference on the exposure, and multiply the first ratio by the first weight to obtain the correction parameter.

[0068] The exposure difference coefficient generation module is configured to multiply the basic exposure difference direction parameter by the correction parameter to generate a double-sided exposure direction factor.

[0069] In the above embodiments, since the DR detector body 1 supports "double-sided exposure", but there are inherent differences in physical characteristics between the two sides, when it is exposed on the front side (outwardly bent to fit the outer wall), the X-ray source is located on the outside of the DR detector body 1, the incident path is short and the vertical component is large, and the energy attenuation is uniform. When it is exposed on the back side (inwardly bent to fit the inner wall), the X-ray source needs to penetrate the DR detector body 1 (such as the support layer and circuit layer) to reach the scintillator. The path is long and the oblique incident component is large, and the energy attenuation is uneven, resulting in the sensitivity of the back side being lower than that of the front side. Moreover, the bending direction (inward / outward bending) directly changes the relative angle between the detector and the X-ray source. The double-sided exposure direction factor calculated by the double-sided exposure compensation module provides an accurate basis for grayscale compensation by quantifying the difference through multi-parameter collaborative quantification.

[0070] In the double-sided exposure compensation module, the formula for calculating the double-sided exposure direction factor is as follows:

[0071] ;

[0072] In the formula: B is the double-sided exposure direction factor, f is the bending direction coefficient (=1 when the inner bend is attached to the outer curved surface, and =-1 when the outer bend is attached to the inner curved surface; user selection or automatic sensor recognition), b is the exposure surface efficiency coefficient (=1.0 for front exposure, 0.9 for back exposure), specifically based on the uniformity test of the Gadox scintillator double-sided coating, QC. ’ max D is the normalized maximum curvature. ref D is the reference distance to the radiation source, and D is the actual distance to the radiation source. It can directly distinguish between two scenarios: "inner bend reverse exposure" (fitting the inner wall of the pipe) and "outer bend front exposure" (fitting the outer wall of the pipe), in conjunction with... Reflects the uniformity of curvature distribution. The exposure dose difference caused by the change in X-ray source distance is corrected, and thus the double-sided exposure direction factor B, which quantifies the difference in physical characteristics during double-sided exposure, is obtained.

[0073] Furthermore, it is worth noting that the bending direction (inward bend / outward bend) and the exposure surface selection (front / back side) are the core links connecting the physical state of the DR detector body 1 with the algorithm correction:

[0074] When the DR detector body 1 bends inward, the concave surface fits against the inner wall; when it bends outward, the convex surface fits against the outer wall. This directly changes the incident angle of the rays (the oblique incident component of the rays is larger when bending inward) and the stress state of the DR detector body 1 (the inner edge is under tension, and the outer edge is under compression). This results in completely different curvature distribution and pixel stretching characteristics. Secondly, when the front side is exposed, the scintillator directly receives the rays, while when the back side is exposed, it needs to penetrate the substrate material. The energy response curves of the two are different. If the exposure surface selection is ignored, the grayscale correction will completely deviate from the physical reality. Therefore, the bending direction is determined by the bending direction coefficient f, and the exposure surface selection is determined by the exposure surface efficiency coefficient b. Together, they ensure that the double-sided exposure direction factor B can accurately quantify the "bending-exposure" coupling effect.

[0075] In summary, this embodiment quantifies the degree of exposure unevenness and clarifies its direction by coupling multiple physical parameters, providing a precise basis for dynamic exposure compensation and effectively improving image grayscale uniformity.

[0076] Example 4:

[0077] Please see Figure 1 and Figure 3 Based on Embodiment 3, the scenario complexity adaptation module provided in this embodiment of the invention specifically includes:

[0078] The curvature dispersion calculation module is configured to collect curvature distribution data through a strain sensor array, calculate the standard deviation of the curvature distribution, and divide the standard deviation by a preset reference curvature dispersion to obtain the normalized curvature dispersion parameter.

[0079] The exposure difference integration module is configured to obtain the absolute value of the double-sided exposure direction factor, multiply it by the normalized value of curvature dispersion and the preset material attenuation coefficient, and generate the basic parameters of scene complexity.

[0080] The acquisition time weighting module is configured to acquire the actual image acquisition time through a timer, calculate the second ratio of the actual image acquisition time to the preset reference acquisition time, and multiply it by a preset weight to obtain the time efficiency parameter.

[0081] The complexity index generation module is configured to add the scene complexity base parameters output by the exposure difference integration module and the acquisition time weighting module to the time efficiency parameter to generate the bending complexity index.

[0082] The mode switching control module is configured to trigger adaptive selection of the detection mode based on the scene complexity index.

[0083] In the above embodiments, the application scenarios of the flexible DR detector are complex and diverse. Different scenarios have conflicting requirements for "imaging accuracy" and "detection efficiency". The scenario complexity adaptation module quantifies the complexity through the bending complexity index and feeds back to correct the curvature parameters in the geometric distortion correction module because: in low-complexity scenarios (such as flat metal parts), detection efficiency should be prioritized; in high-complexity scenarios (such as composite material curved surface welding), imaging accuracy should be prioritized (long acquisition time, multiple algorithm correction). If the bending complexity index is not quantified, the algorithm will not be able to automatically match the strategy, resulting in "overcorrection" or "undercorrection". The higher the bending complexity index, the more uneven the curvature distribution of the scenario and the more complex the material attenuation. At this time, it is necessary to increase the sensitivity of the geometric distortion correction module to "extreme curvature" to avoid distortion correction omissions due to insufficient curvature sampling. Conversely, in low bending complexity index scenarios, reducing the curvature weight can reduce redundant calculations and improve efficiency.

[0084] In the scene complexity adaptation module, the formula for calculating the bending complexity index is as follows:

[0085] ;

[0086] In the formula: WF is the bending complexity index, σ QC σ is the standard deviation of curvature calculated in real time using data from the strain sensor array of the DR detector body 1. QC-ref The reference curvature dispersion is denoted by s, which is the material attenuation coefficient, set according to the material of the object being detected (e.g., metal = 1.0, composite material = 1.5), and T is the actual image acquisition time. ref For reference collection time, among which, This collectively reflects the "difficulty in imaging". The data collection time is converted into a complexity contribution.

[0087] Furthermore, the bending complexity index (WF) is essentially a quantitative mapping of the physical characteristics of the scene. Its value directly reflects the matching relationship between "imaging difficulty" and "inspection requirements." Through adaptive selection of detection modes: when the bending complexity index < 5, such scenes (e.g., flat steel structures) have low sensitivity to distortion, and the high-speed mode can improve detection efficiency while maintaining a high defect recognition rate, so the high-speed mode is selected; when the bending complexity index 5 ≤ bending complexity index < 8, the selected hybrid mode can avoid "black and white" mode switching, and achieve a more efficient detection purpose by balancing efficiency and accuracy; when the bending complexity index ≥ 8, such scenes (e.g., carbon fiber curved surface welding) have low defect contrast, and the handheld high-precision mode can improve the defect recognition rate, meet the inspection standard requirements, and thus achieve a balance between the engineering contradiction of "inspection efficiency" and "imaging accuracy."

[0088] In summary, this embodiment quantifies the complexity of the scene through multi-dimensional parameters, realizes intelligent switching of detection modes, balances detection efficiency and accuracy, and adapts to complex industrial detection needs.

[0089] Example 5:

[0090] Please see Figure 3 Based on Embodiment 4, the control module provided in this embodiment of the invention performs curvature parameter correction, including correction of the maximum curvature and the minimum curvature. The specific correction formula is as follows:

[0091] Q corr max =QC max ×(1+WF×0.02);

[0092] Q corr min =QC min ×(1-WF×0.02);

[0093] In the formula: Q corr max Q is the adjusted maximum curvature. corr min To achieve the minimum curvature after adjustment, QC max For maximum curvature, QC min WF is the minimum curvature, and WF is the bending complexity index.

[0094] In the above embodiments, the bending complexity index WF is calculated using the modified geometric distortion correction module formula. Maximum curvature QC max and minimum curvature QC min The weights are adjusted to achieve "dynamic sensitivity enhancement" for complex scenarios. The higher the bending complexity index (WF) (the more complex the scenario), the higher the maximum curvature QC. max Magnified, minimum curvature QC min The reduction in size leads to a larger overall distortion index Q. diff The dynamic range is expanded, thereby enhancing the algorithm's ability to perceive minute curvature changes (such as local curvature fluctuations caused by pores inside composite materials). 0.02 is the optimized value for the experimental example, and can be flexibly set according to the specific detection conditions of the DR detector body 1.

[0095] In summary, this embodiment further optimizes the accuracy of geometric correction by introducing the bending complexity index WF to dynamically adjust the curvature, and performs particularly well in complex bending scenarios.

[0096] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A flexible DR detector, characterized in that, The DR detector body (1) includes an edge frame (2), a flexible detection plate (3), and a stabilizing wheel (6) which are fixedly connected to the DR detector body (1). The processor inside the DR detector body (1) is fixedly connected to a geometric distortion correction module, a double-sided exposure compensation module, a scene complexity adaptation module, and a control module. The geometric distortion correction module is configured to collect the curvature and pixel stretching data of the flexible probe plate (3) when it is bent, and calculate and generate a comprehensive distortion index by quantifying the coupling effect of macroscopic bending difference and microscopic pixel deformation and triggering regional correction. The double-sided exposure compensation module is configured to receive the comprehensive distortion index generated by the geometric distortion correction module, and couple the bending direction, exposure surface posture and X-ray source distance parameters of the flexible detection plate (3) during detection, quantify the non-uniformity of the double-sided exposure of the flexible detection plate (3), and calculate and generate the double-sided exposure direction factor. The scene complexity adaptation module is configured to receive the double-sided exposure direction factor generated by the double-sided exposure compensation module, integrate the curvature dispersion and acquisition time dimension data of the flexible detection plate (3) during detection, and combine the material attenuation of the detected material to quantify the scene complexity, calculate and generate the bending complexity index, and adaptively switch the detection mode according to the bending complexity index. The control module is configured to receive parameters calculated and generated by the geometric distortion correction module, the double-sided exposure compensation module and the scene complexity adaptation module, as well as the switched detection modes, and control the DR detector body (1) to perform curvature parameter correction and detection mode switching.

2. The flexible DR detector according to claim 1, characterized in that, Both sides of the flexible detection plate (3) are fixedly connected to edge frames (2), one side of the edge frame (2) is fixedly connected to a flexible handle (4), and the other two sides of the edge frame (2) away from the flexible handle (4) are fixedly connected to strapping hole frames (5). The stabilizing wheel (6) is fixedly connected to the inside of the strapping hole frame (5) by a strapping strap fixing assembly.

3. A flexible DR detector according to claim 1, characterized in that: The geometric distortion correction module specifically includes: The macro curvature difference calculation module is configured to collect the maximum curvature and minimum curvature of the flexible probe plate (3) when it is bent by the strain sensor array distributed on the substrate surface of the flexible probe plate (3), calculate the first difference between the maximum curvature and the minimum curvature, and divide the first difference by the preset reference curvature to obtain the normalized macro curvature difference parameter. The micro-pixel stretching calculation module is configured to measure the axial length change of the DR detector body (1) by means of a grating ruler fixedly connected to the DR detector body (1), and multiply the axial length change by a preset original pixel spacing, and then divide by the square of the reference length of the DR detector body (1) in the square state to obtain the micro-pixel stretching parameters. The comprehensive distortion index generation module is configured to add the macro curvature difference parameter and the micro pixel stretching parameter to generate the comprehensive distortion index, and to divide the high-precision correction area and the high-speed correction area according to the comprehensive distortion index.

4. A flexible DR detector according to claim 3, characterized in that: The double-sided exposure compensation module specifically includes: The bending direction and exposure surface coupling module is configured to determine the bending direction of the flexible detection plate (3) during detection by a Hall element, and assign a preset bending direction coefficient according to the bending direction. The exposure surface of the flexible detection plate (3) during detection is determined by an attitude sensor, and a preset exposure surface efficiency coefficient is assigned to the exposure surface. The two coefficients are multiplied by the exposure surface efficiency coefficient to obtain the basic exposure difference direction parameter. The X-ray source distance correction module is configured to acquire the actual X-ray source distance through a laser rangefinder, calculate the first ratio between the actual X-ray source distance and the preset reference X-ray source distance, divide the comprehensive distortion index by the normalized maximum curvature to obtain the first weight of the curvature difference on the exposure, and multiply the first ratio by the first weight to obtain the correction parameter. The exposure difference coefficient generation module is configured to multiply the basic exposure difference direction parameter by the correction parameter to generate the double-sided exposure direction factor.

5. A flexible DR detector according to claim 1, characterized in that: The scenario complexity adaptation module specifically includes: The curvature dispersion calculation module is configured to collect curvature distribution data through a strain sensor array, calculate the standard deviation of the curvature distribution, and divide the standard deviation by a preset reference curvature dispersion to obtain the normalized curvature dispersion parameter. The exposure difference integration module is configured to obtain the absolute value of the double-sided exposure direction factor, and multiply it by the curvature dispersion normalization value and the preset material attenuation coefficient to generate the basic parameters of scene complexity. The acquisition time weighting module is configured to acquire the actual image acquisition time through a timer, calculate the second ratio of the actual image acquisition time to the preset reference acquisition time, and multiply it by a preset weight to obtain the time efficiency parameter. The complexity index generation module is configured to add the scene complexity base parameters output by the exposure difference integration module and the acquisition time weighting module to the time efficiency parameter to generate the bending complexity index. The mode switching control module is configured to trigger adaptive selection of the detection mode based on the scene complexity index.

6. A flexible DR detector according to claim 5, characterized in that: The adaptive selection based on the detection mode includes: If the bending complexity index is less than 5, then the high-speed mode is selected; If 5 ≤ the bending complexity index < 8, then the hybrid mode is selected; If the bending complexity index is ≥8, then the high-precision mode is selected.

7. A flexible DR detector according to claim 3, characterized in that: The curvature parameter correction performed by the control module includes the correction of the maximum and minimum curvature, and the specific correction formula is as follows: Q corr max =QC max ×(1+WF×0.02); Q corr min =QC min ×(1-WF×0.02); In the formula: Q corr max Q is the adjusted maximum curvature. corr min To achieve the minimum curvature after adjustment, QC max For maximum curvature, QC min WF is the minimum curvature, and WF is the bending complexity index.

8. A flexible DR detector according to claim 4, characterized in that: The bending direction coefficient is set according to the outward and inward bending directions, and the bending direction coefficient is -1 when bending outward and 1 when bending inward. The exposure surface coefficient is set according to the front and back of the exposure surface. When it is the front, the exposure surface coefficient is set to 1, and when it is the back, the exposure surface coefficient is set to 0.9.

Images