X-ray digital imaging automatic detection system for multi-split transmission conductor strain clamp

Abstract

The application discloses a multi-split power transmission conductor strain clamp X-ray digital imaging automatic detection system and relates to the field of electric power equipment operation and inspection. The system comprises: a gate planning module, which collects radiation states and stability to obtain state codes, generates session numbers, view sequence numbers and session records; a calibration collection module, which generates external parameter numbers and exposure numbers for the state codes in the permission state, collects frame sequence numbers for the view sequence numbers, and generates collection records; a pretreatment correction module, which reads the collection records to obtain compensation numbers, external parameter numbers and frame sequence numbers, generates standardized images, structure indexes and pretreatment records; a defect judgment module, which obtains structure indexes and standardized images based on the pretreatment records, generates defect types and evidence references, and writes into judgment records; and an archiving and tracing module, which reads the session records, collection records, pretreatment records and judgment records, associates session numbers and state codes, and writes into archiving records. The gate session is connected through collection, pretreatment, judgment and archiving to realize full-chain tracing.

Description

Technical Field

[0001] This invention relates to the field of power equipment operation and maintenance, specifically to an automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines. Background Technology

[0002] Tension clamps for multi-split transmission conductors are key load-bearing hardware in transmission lines, responsible for conductor tension transmission, clamping and positioning, and handling coupled loads from long-term service environments. They are commonly found in long spans, tension sections, and transposition zones. These clamps typically include a crimping tube, a wedge-shaped crimping structure, a crimping transition section, and connecting hardware. Their internal contact interfaces, crimping tightness, and structural integrity directly affect conductor gripping force, heat generation level, and fatigue life. Due to manufacturing deviations, inconsistent on-site crimping processes, and long-term wind vibration and icing load disturbances, hidden defects such as crimping loosening, porosity, localized cracks, interface debonding, and foreign matter inclusions may occur inside the clamps. These defects are characterized by inconspicuous appearance, slow development, and serious consequences. If these defects continue to evolve, they may lead to increased contact resistance, abnormal localized temperature rise, hardware breakage, or conductor slippage, posing a threat to the safe operation of the line.

[0003] For internal defects in wire clamps, X-ray digital imaging has become an effective non-destructive testing method due to its penetrating observation capability. However, the structure of multi-split conductor tension clamps is complex. The number and spatial arrangement of conductor splits cause overlapping obstructions in the X-ray penetration path. The thickness variation and scattering effect of the fitting material are significant, and the imaging grayscale response is easily affected by exposure level, geometric extrinsic parameters, detector response nonlinearity, and fixed-mode artifacts. In the field inspection environment, there are also radiation safety constraints and personnel intrusion risks. Frame vibration, tooling micro-displacement, and shooting posture deviations further cause blurring and geometric mismatch, making the same defect appear unstable from different perspectives. Traditional manual interpretation processes suffer from insufficient consistency, untraceable evidence citation, and difficulty in verifying judgment results in high-throughput inspection tasks. Image processing methods that only use single filtering or simple threshold segmentation often fail to simultaneously achieve artifact suppression and defect boundary fidelity, and it is difficult to establish an interpretable evidence chain at the structural segment level.

[0004] Meanwhile, the operation and maintenance of transmission lines places higher demands on the engineering management of testing data: testing tasks need to solidify radiation states and stability conditions on a session-by-session basis; the acquisition phase needs to form a consistent recording link between external parameter information, exposure information, and frame sequences; the preprocessing phase needs to strictly correspond compensation configurations with external parameter versions; the judgment phase needs to output defect types, location indexes, and evidence references to support verification; and the archiving phase needs to associate session records, acquisition records, preprocessing records, and judgment records and write them into the archived records to achieve traceable, auditable, and searchable end-to-end management. In existing technologies, common solutions do not provide sufficiently detailed descriptions of the data associations between gating planning, calibration acquisition, preprocessing correction, defect judgment, and archiving traceability. They lack a recording system and consistency identification mechanism oriented towards engineering sites, resulting in an unstable mapping relationship between test results and original evidence, making it difficult to meet the comprehensive requirements of automation, traceability, and consistency in large-scale operation and maintenance scenarios. Therefore, it is necessary to build an automatic detection system for X-ray digital imaging of multi-splitter tension clamps, and establish a clear data link and recording system between radiation safety gating, calibration acquisition, calibration consistency, structural constraint determination and archiving traceability, so as to support stable and reliable engineering applications. Summary of the Invention

[0005] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide an automatic X-ray digital imaging detection system for tension clamps of multi-split transmission lines, so as to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines, comprising: The gating planning module collects radiation status and stability to obtain status codes, and generates session numbers, view sequence numbers and session records. The calibration acquisition module reads the session record. If the status code is in the permitted state, it generates the external parameter number, exposure number, view sequence number, acquisition frame sequence number, and generates the acquisition record. The preprocessing and correction module reads the acquisition records to obtain the compensation number, external parameter number, and frame sequence number, and generates standardized images, structural indexes, and preprocessing records. The defect determination module obtains structural indexes and standardized images based on preprocessed records, generates defect types and evidence references, and writes them into the determination record. The archive traceability module reads session records, collection records, preprocessing records, and judgment records, and writes them into the archive records along with the session number and status code.

[0007] The present invention is further configured such that the gating planning module includes: The dose rate sequence, interlocking markers, and source preparation markers are collected to form a radiation state record, and the displacement sequence and acceleration amplitude sequence are collected to form a stability record. Based on the upper limit of the measurement range, normalized sequences are generated from the dose rate sequence, displacement sequence, and acceleration amplitude sequence and written into the gated intermediate record; A radiation compliance index is generated based on the normalized sequence and dose limit; a stability index is generated based on the normalized sequence and displacement and vibration limits; and a gating discrimination quantity is generated based on the radiation compliance index and the stability index. Status codes are generated from gating discriminants based on radiation threshold, stability threshold, and discrimination threshold. A session number is generated based on the task identifier, device identifier, time count, and status code. A view sequence number is generated based on the view index sequence. The session number, view sequence number, and status code are written into the session record.

[0008] The present invention is further configured such that the calibration acquisition module includes: The calibration acquisition module reads the session record to obtain the session number, view sequence number, and status code; When the status code is in the permitted state, a view index sequence is formed based on the view sequence number, and an association identifier is established between the view index sequence and the session number; Based on the viewpoint index sequence, set the calibration posture and collect calibration frames, then establish an association between the calibration frames and the viewpoint index; Based on the calibration frame, target feature coordinates and target feature confidence values ​​are extracted, and the target feature coordinates and target feature confidence values ​​are associated with the session number and view index.

[0009] The present invention is further configured such that the extrinsic exposure frame sequence record includes: The calibration acquisition module generates a projection coefficient set based on the target spatial coordinates, target feature coordinates and target feature confidence, generates an external parameter number based on the projection coefficient set, and writes the external parameter number, session number and view sequence number into the acquisition record. An exposure number is generated based on the set of exposure levels and the upper limit of the dose. The exposure number, session number, and external parameter number are written into the acquisition record. Frame sequence numbers are generated by collecting frame numbers based on the viewpoint index sequence, and the frame sequence number, viewpoint sequence number, and exposure number are written into the acquisition record. A consistency marker is generated based on the session number, external parameter number, exposure number, frame sequence number, and status code. The consistency marker is written into the acquisition record and associated with the status code.

[0010] The present invention is further configured such that the preprocessing correction module includes: The preprocessing correction module reads the acquisition record to obtain the compensation number, external parameter number, and frame sequence number, dereferences the frame sequence number to obtain the frame number set and the corresponding original image, and establishes a relationship between the frame number set and the frame sequence number. A frame sequence consistency marker is generated based on the frame number set, and the frame sequence consistency marker is associated with the compensation number, external parameter number, and frame sequence number. Based on the dereference of the compensation number, the dark field, flat field, bad pixel mask, scattering field baseline, and response mapping table are obtained to form a correction configuration group, and the correction configuration group is associated with the compensation number. Based on the correction configuration group, the original image is subjected to response mapping, flat dark field restoration and scattering baseline correction to generate a corrected image, and the corrected image is associated with the frame number set.

[0011] The present invention is further configured such that the artifact suppression and structure indexing include: The preprocessing correction module generates a bad pixel replacement image for the corrected image based on the bad pixel mask, and establishes a relationship between the bad pixel replacement image and the frame number set. Based on the bad pixel replacement image, a fixed pattern term and a structural response term are generated to form an artifact separation image, and the artifact separation image is associated with the compensation number. Based on artifact separation, standardized images and quality labels are generated, and the standardized images and quality labels are associated with frame sequence numbers. The projection coefficient set is obtained by dereferencing the extrinsic parameter number. A structural index is generated for the standardized image. The standardized image, structural index, quality identifier, compensation number, extrinsic parameter number, and frame sequence number are written into the preprocessing record and a preprocessing consistency mark is generated.

[0012] The present invention is further configured such that the defect determination module includes: The defect determination module reads the preprocessing record to obtain standardized images, structural indexes, and frame numbers, generates frame alignment identifiers, and establishes a correlation with the preprocessing record. A detection domain indicator is generated based on the structure index and the structure segment set identifier, and the detection domain indicator is associated with the corresponding frame number. A structural constraint saliency map is generated based on standardized images and detection domain indicators, and the structural constraint saliency map is associated with the corresponding frame number. Candidate domains and candidate domain location indices are generated based on the structural constraint saliency map, and the candidate domains, candidate domain location indices, and structural indexes are associated and written into candidate records.

[0013] The present invention is further configured such that the segment evidence discrimination includes: The defect determination module reads candidate records to obtain candidate domain position indexes, generates structural segment-level evidence based on the structural constraint saliency map according to the structural index, and establishes a correlation between the structural segment-level evidence and the frame alignment identifier. Based on the amount of evidence at the structural segment level, a cross-frame consistency factor is generated, and the cross-frame consistency factor is associated with the structural segment set identifier. Defect type identifiers are generated based on structural segment-level evidence quantity, cross-frame consistency factor, and defect type fingerprint parameters, and the defect type identifiers are associated with structural segment set identifiers. Evidence references and location indexes are generated based on session number, frame alignment identifier, candidate domain location index, and defect type identifier. The defect type identifier, location index identifier, and evidence reference are written into the judgment record and associated with the preprocessing record.

[0014] The present invention is further configured such that the defect determination module includes: Based on the judgment record, the defect type identifier, location index identifier, and evidence citation identifier are obtained to form an archived field set and associate it with the session number; A session association key is generated based on the session number, status code, and view sequence number; an acquisition association key is generated based on the session number, external parameter number, exposure number, and frame sequence number; and a preprocessing association key is generated based on the session number, compensation number, standardized image identifier, and structure index identifier. The three types of association keys are associated with the archive field set. An evidence mapping index is generated based on the evidence citation identifier, location index identifier, defect type identifier, and frame sequence number. The evidence mapping index is then associated with the standardized image identifier and the structural index identifier. Generate a set of traceability edges based on the archived field set, generate a graph signature, and associate the graph signature with the archived field set.

[0015] The present invention is further configured such that the archive generation index includes: The archiving and tracing module generates an archive list based on the session association key, the collection association key, the preprocessing association key, and the image signature and evidence mapping index. It then generates an archive number based on the archive list and associates it with the session number. A partition index is generated based on the device identifier and time count. A path index is generated based on the partition index. The partition index and the path index are associated with the archive number. Generate an archive consistency tag based on the archive number, session number, status code, path index, and evidence citation identifier, and associate the archive consistency tag with the archive number; Archive records are generated based on archive number, session number, and archive consistency marker, and an association is established between session number and status code index.

[0016] This invention provides an automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors. The system comprises: a task modeling module that generates imaging constraints and exposure parameters based on the number of splits, conductor spacing, and clamp type; a geometric calibration and interlocking module that generates extrinsic parameters based on ranging data, and outputs exposure permits based on radiation state, intrusion results, and stability; an imaging acquisition and preprocessing module that acquires frames based on imaging constraints, extrinsic parameters, and exposure permits, obtains an acquired image through registration and fusion, and forms a standard image through dark-field and bright-field reference correction; a structure identification and defect discrimination module that identifies the location, defect type, and defect range based on the standard image; and a self-consistent loop and archiving module that constructs a self-consistent index based on geometric, motion, scattering, coverage, and uncertainty evidence. The self-consistent index updates the exposure parameters or target pose, and a report is generated by writing the task identifier into the standard image, defect results, and self-consistent index. The beneficial effects include: 1. Session numbers and status codes are used throughout the entire process of gating, data acquisition, preprocessing, judgment, and archiving, forming a unified link from session records, acquisition records, preprocessing records, judgment records to archiving records. External parameter numbers, exposure numbers, frame sequence numbers, compensation numbers, structural index identifiers, and evidence citation identifiers are all associated within the same link, ensuring a clear traceability path and verifiable record elements; 2. Gated planning drives calibration acquisition under permissive constraints. The view sequence number forms a unified organizational relationship between the acquisition posture and the frame sequence number. The extrinsic parameter number and exposure number are written into the acquisition record and associated with the session number. The acquisition record provides a stable correspondence between the compensation number, extrinsic parameter number, and frame sequence number for preprocessing correction, reducing the disturbance of acquisition condition differences to subsequent processing links. 3. Preprocessing and correction outputs standardized images and structural indexes, which are written to the preprocessing record. Defect judgment generates defect type, location index, and evidence references under the constraints of the structural index and writes them to the judgment record. Archiving and traceability solidify the association relationship through session association keys, acquisition association keys, preprocessing association keys, evidence mapping indexes, and image signatures, achieving consistent referencing and retrieval between the judgment conclusion and the original frame sequence and processing results.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating an exemplary embodiment of the present invention of an automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines. Detailed Implementation

[0019] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0021] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0022] Example 1:

[0023] Automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors, such as... Figure 1 As shown, it includes: The gating planning module collects radiation status and stability to obtain status codes, and generates session numbers, view sequence numbers and session records. The calibration acquisition module reads the session record. If the status code is in the permitted state, it generates the external parameter number, exposure number, view sequence number, acquisition frame sequence number, and generates the acquisition record. The preprocessing and correction module reads the acquisition records to obtain the compensation number, external parameter number, and frame sequence number, and generates standardized images, structural indexes, and preprocessing records. The defect determination module obtains structural indexes and standardized images based on preprocessed records, generates defect types and evidence references, and writes them into the determination record. The archive traceability module reads session records, collection records, preprocessing records, and judgment records, and writes them into the archive records along with the session number and status code.

[0024] The present invention is further configured such that the gating planning module includes: The radiation status record is formed by collecting dose rate sequences, interlock markers, and source preparation markers, while the stability record is formed by collecting displacement sequences and acceleration amplitude sequences. Specifically, at the start of the detection task, a gated time window is opened for data acquisition. The dose rate monitor continuously reads dose rate values ​​at a fixed sampling period to form a timestamped dose rate sequence. Simultaneously, the valid markers of the interlocking loops and the radiation source preparation markers are read as discrete elements of the radiation safety status and written into the radiation status record along with the dose rate sequence. The displacement sequence from the displacement sensor and the acceleration amplitude sequence from the acceleration sensor are acquired in parallel. The displacement sequence reflects the tooling attitude drift trend, and the acceleration amplitude sequence reflects vibration intensity fluctuations. Both are aligned with the same timestamp and written into the stability record. Based on the upper limit of the measurement range, normalized sequences are generated from the dose rate sequence, displacement sequence, and acceleration amplitude sequence and written into the gated intermediate record. Specifically, the upper limits of the dose rate range, displacement range, and acceleration range are read from the device configuration table. Each sampled value in the dose rate sequence is scaled proportionally according to the upper limit of the dose rate range to make it fall into a uniform dimensionless range. The displacement sequence is first taken as the absolute value to eliminate the influence of direction, and then scaled proportionally according to the upper limit of the displacement range. The acceleration amplitude sequence is scaled proportionally according to the upper limit of the acceleration range. The three scaled sequences and their corresponding timestamps are written into the gated intermediate record, and the original sequences are retained in the record index field to ensure subsequent traceability. A radiation compliance index is generated based on a normalized sequence and dose limits, and a stability index is generated based on a normalized sequence and displacement and vibration limits. A gating discrimination quantity is generated based on the radiation compliance index and the stability index. Specifically, dose limits, displacement limits, and vibration limits are read from a threshold configuration table. The dose limits are compared point by point on the normalized dose rate sequence. The portion exceeding the limit is accumulated according to the principle of "the larger the magnitude of the exceedance and the longer the duration, the heavier the penalty". The accumulated result is then mapped to the radiation compliance index, so that short-term slight fluctuations have little impact on the index, while continuous exceedances have a significant impact on the index. At the same time, the interlocking effective marker and the source preparation marker are necessary prerequisites for the radiation compliance index. If either marker is abnormal, the radiation compliance index will be directly placed in the low value range. A status code is generated from the gating discrimination quantity based on the radiation threshold, stability threshold, and discrimination threshold. Specifically, the gating policy table reads the radiation threshold, stability threshold, and discrimination threshold, compares the radiation compliance index with the radiation threshold, compares the stability index with the stability threshold, and simultaneously compares the gating discrimination quantity with the discrimination threshold. If all three conditions are met, the status code is set to the permitted state; if the radiation condition is met but the stability condition is not, the status code is set to the pending stability state; if the radiation condition is not met, the status code is set to the rejected state. The status code and the start and end times of the gating time window are written into the gating intermediate record as a reference source for the generation of subsequent session records. A session number is generated based on the task identifier, device identifier, time count, and status code. A view sequence number is generated based on the view index sequence. The session number, view sequence number, and status code are written into the session record. Specifically, the task identifier, device identifier, and time count are read, and the status code is used as a participating element. The session number is obtained by encoding and combining these elements according to the rule that "the same task and the same device generate different session numbers at different times, and changes in the status code will change the session number." The view index sequence configuration is read, and the view index sequence is serialized and encoded in a predetermined order to obtain the view sequence number, making the view sequence number sensitive to view order and reproducible. The session number, view sequence number, and status code are written into the session record, and a reference identifier for the gating intermediate record is saved in the session record, ensuring that the subsequent calibration and acquisition module can trace the gating basis and the source of the permission state judgment through the session record.

[0025] The present invention is further configured such that the calibration acquisition module includes: The calibration acquisition module reads session records to obtain the session number, view sequence number, and status code. Specifically, during the calibration acquisition phase, it reads the session number, view sequence number, and status code from the session records, while also reading the start and end information of the time window and the device identification verification field registered in the session records. The status code is then judged to be in a permissible state; if the permissible state is satisfied, the subsequent calibration process begins; if the status code does not satisfy the permissible state, the calibration acquisition action is stopped, and only the log entry and session number reference for this reading behavior are retained to ensure that the cause of the failure to perform calibration can be located later. When the status code is in the permitted state, a view index sequence is formed based on the view sequence number, and an association identifier is established between the view index sequence and the session number. Specifically, the view arrangement rules corresponding to the view sequence number are read from the view library, and the view sequence number is decoded into an ordered view index sequence according to the rules. The decoding process adopts the principle of "taking values ​​within the valid range of the view library, maintaining the view order, and reproducibly corresponding to the session number": first, the view sequence number is mapped to the valid index range of the view library, then it is expanded sequentially according to a preset step size to obtain several view indices, and finally, the expansion results are corrected for out-of-bounds errors and duplicate conflicts are resolved, so that each view index points to a unique view entry in the view library. The view index sequence and the session number are written into an association identifier table, which contains the session number, view sequence number, view index sequence version number, and generation timestamp, as the basis for subsequent attitude setting and calibration frame acquisition. The calibration attitude is set and calibration frames are acquired based on the viewpoint index sequence. A correlation identifier is established between the calibration frames and the viewpoint indexes. Specifically, the calibration attitude is set sequentially item by item in the viewpoint index sequence. Each viewpoint index corresponds to a set of frame attitude parameters and tooling positioning parameters. The attitude setting adopts a "coarse positioning followed by fine alignment" process: first, the rotation and translation mechanisms are driven to the vicinity of the target attitude; then, limit sensors and alignment marks are used for feedback to perform fine adjustments, ensuring that the attitude deviation falls within the attitude tolerance range. After the attitude stabilizes, one or more calibration frames are acquired. During acquisition, the exposure configuration reference, detector operating status, and frame sequence number are recorded simultaneously. A correlation identifier is established between each calibration frame and the corresponding viewpoint index. The correlation identifier includes the session number, viewpoint index, calibration frame sequence number, acquisition timestamp, and original data storage reference, ensuring that each frame of calibration data can be traced back to a specific viewpoint and session. Based on the calibration frame, target feature coordinates and confidence values ​​are extracted, and these coordinates and confidence values ​​are linked to the session number and viewpoint index. Specifically, morphological consistency checks are performed on candidate target regions, eliminating regions that do not meet the target's geometric constraints. Then, target feature coordinates are extracted from the validated target regions, recorded in pixel coordinate form. The extraction process follows the principle of "strong edge response, clear structural abrupt changes, and spatial distribution satisfying the target's geometric constraints": first, edge enhancement and corner response calculations are performed on the target region; then, effective feature points are selected from the response peak points according to the minimum spacing and geometric arrangement rules, resulting in a set of target feature coordinates. The target feature confidence value is composed of three types of evidence: matching score, consistency residual, and local sharpness. The matching score reflects the template fit, the consistency residual reflects the relative geometric deviation of feature points, and the local sharpness reflects the texture discernibility of the feature point's neighborhood. These three types of evidence are mapped to a single confidence value according to threshold and penalty rules, significantly reducing the confidence value when evidence is insufficient or the residual is too large.

[0026] The present invention is further configured such that the extrinsic exposure frame sequence record includes: The calibration acquisition module generates a projection coefficient set based on the target spatial coordinates, target feature coordinates, and target feature confidence values. It then generates extrinsic parameter numbers based on these projection coefficient sets and writes the extrinsic parameter numbers, along with the session number and viewpoint sequence number, into the acquisition record. Specifically, it reads the target spatial coordinate table from the target file, which provides the spatial position of each target feature in the target coordinate system. It also reads the target feature coordinates and target feature confidence values ​​corresponding to each viewpoint index under the same session number from the feature record. A pairing list is established between spatial coordinates and pixel coordinates according to the rule of "one-to-one correspondence between corresponding feature points." The projection relationship is then solved on the pairing list: first, the pairing list is sorted according to the confidence value, prioritizing feature points with higher confidence values ​​for initial solution; after obtaining a set of projection coefficients from the initial solution, the projection residuals are evaluated point by point through back substitution. Low-confidence points with residuals exceeding the allowable range are removed, and the remaining points are used to re-solve and iteratively converge to a stable solution, resulting in the projection coefficient set. Subsequently, the projection coefficient set is version-coded to generate extrinsic parameter numbers. The coding rule adopts the principle that "the same projection coefficient set yields the same extrinsic parameter number under the same coding parameters, and different projection coefficient sets produce different extrinsic parameter numbers." The value range of the projection coefficients is segmented, quantized, and combined into extrinsic parameter numbers. The session number, view sequence number, extrinsic parameter number, and projection coefficient set reference identifier are written into the acquisition record, and references to feature records and target files are retained in the acquisition record to ensure that the extrinsic parameter number can be traced back to the original observation element. An exposure number is generated based on the exposure level set and dose limit. This exposure number, along with the session number and external parameter number, is written into the acquisition record. Specifically, the equipment file reads the exposure level set, which contains several combinations of voltage, current, and exposure duration levels. Simultaneously, it reads the dose limit for this task and the radiation status record referenced during the gating planning phase. A feasibility assessment is performed on each exposure level combination: first, the dose contribution corresponding to the radiation intensity per unit time is estimated based on the level combination; then, the dose consumption of that level combination in a single frame acquisition is calculated based on the exposure duration. The calculated value is compared with the dose limit. Level combinations exceeding the dose limit are directly marked as unusable. Level combinations meeting the dose constraints are further refined based on the detector dynamic range and expected penetration thickness, prioritizing combinations that avoid saturation and have sufficient contrast. The finally selected level combinations generate exposure numbers according to the rule of "level parameter quantization encoding plus session association encoding," making the exposure number sensitive to changes in the level combination while establishing a stable association with the session number and external parameter number. The exposure number, session number, and external parameter number are written together into the acquisition record, along with the exposure level reference identifier and dose upper limit reference identifier, to ensure that the source of the exposure number can be verified. Frame sequence numbers are generated based on the viewpoint index sequence. These frame sequence numbers, along with the viewpoint sequence number and exposure number, are written into the acquisition record. Specifically, the viewpoint index sequence and exposure number are used to perform formal imaging acquisition in the order of the viewpoint index sequence. Each viewpoint index corresponds to one attitude setting and one set of frame acquisition actions. Frame numbers are generated using a fixed frame rate or fixed trigger interval during frame acquisition. Each frame number contains three parts: a session number reference, a viewpoint index reference, and an intra-frame sequence number, ensuring the frame number is unique within the same session. The frame numbers acquired from each viewpoint are arranged into a frame number list according to the acquisition order, and a frame sequence number is generated from this list. The frame sequence number generation process uses a "sequence-sensitive combined encoding" rule: the frame number list is first quantized and mapped item by item, and then concatenated and encoded sequentially to obtain a single sequence number. This ensures that any omission, insertion, or rearrangement of frame numbers will result in a change in the frame sequence number. The frame sequence number, along with the viewpoint sequence number and exposure number, is written into the acquisition record. Simultaneously, a frame number list reference identifier and an original image storage reference identifier are also written, providing an index entry for the subsequent preprocessing and correction module to dereference the original image by frame sequence number. A consistency marker is generated based on the session number, external parameter number, exposure number, frame sequence number, and status code. This consistency marker is written to the acquisition record and associated with the status code. Specifically, the session number, external parameter number, exposure number, frame sequence number, and status code are read to generate the consistency marker. The consistency marker uses a "link element binding encoding" rule, with the session number as the primary key, the external parameter number, exposure number, and frame sequence number as key version elements of the acquisition link, and the status code as a permissive constraint element. These elements are combined in a preset encoding order to generate a single marker value, ensuring that any change in any element triggers a change in the consistency marker. The consistency marker is written to the acquisition record, and an association between the consistency marker and the status code is established. This association simultaneously points to the status code source entry and the gated time window reference entry in the session record. At this point, a closed reference chain is formed within the acquisition record: "session number—view sequence number—external parameter number—exposure number—frame sequence number—consistency marker—status code," satisfying the version consistency requirements of the compensation number, external parameter number, and frame sequence number for subsequent preprocessing correction.

[0027] The present invention is further configured such that the preprocessing correction module includes: The preprocessing correction module reads the acquisition records to obtain the compensation number, extrinsic parameter number, and frame sequence number. It then dereferences the frame sequence number to obtain a set of frame numbers and the corresponding original image, establishing a correlation between the frame number set and the frame sequence number. Specifically, during the preprocessing correction stage, the module reads the compensation number, extrinsic parameter number, and frame sequence number from the acquisition records, along with the frame number list references and original image storage references registered in the acquisition records. By dereferencing the frame sequence number in the frame index table, it obtains a set of frame numbers arranged in the acquisition order. Based on this set of frame numbers, it locates the original image files frame by frame, forming a set of original images that corresponds one-to-one with each frame number. The frame number set and frame sequence number are then correlated and written into a frame association table. This frame association table includes the frame sequence number, a summary of the frame number set, a frame count, and a reference to the original image storage location, ensuring that subsequent processing is carried out within the same frame set and traceable to specific frames. A frame sequence consistency marker is generated based on the frame number set, and then associated with the compensation number, extrinsic parameter number, and frame sequence number. Specifically, a consistency generation process is performed on the frame number set to reflect the completeness and order consistency of the frame set. The process is as follows: frame numbers are read item by item in the order of the frame number set, and each frame number is mapped to a fixed-length coded segment; the coded segments are concatenated in sequence to form a set digest, and then the digest is compressed and encoded to obtain the frame sequence consistency marker. During the generation process, the frame numbers are checked for duplicates, missing numbers, or broken sequences. If any abnormality is found, an abnormality type flag and an abnormality position index are recorded in the consistency marker's auxiliary field. The frame sequence consistency marker is associated with the compensation number, extrinsic parameter number, and frame sequence number and written into the consistency record, so that subsequent correction results can be traced back to the specific compensation version and specific frame set. Based on the dereference of the compensation number, dark field, flat field, bad pixel mask, scattered field baseline, and response mapping table are obtained, forming a correction configuration group. The correction configuration group is then associated with the compensation number. Specifically, the compensation number is dereferenced in the compensation configuration library to obtain the dark field data, flat field data, bad pixel mask, scattered field baseline, and response mapping table required for this correction. Dark field data is used to characterize the detector's background response under no-illumination conditions; flat field data is used to characterize pixel gain differences under uniform illumination conditions; bad pixel mask is used to mark the locations of failed or abnormal pixels; scattered field baseline is used to characterize the fixed scattered background distribution; and the response mapping table is used to describe the nonlinear response of the detector and the mapping relationship between quantized grayscale and the radiation response domain. These five types of data are packaged into a correction configuration group according to the same version. The correction configuration group includes the compensation number, configuration version number, generation timestamp, effective resolution range, and data integrity check code. The correction configuration group is associated with the compensation number and written into the configuration association table to ensure that the same compensation number can reproduce a consistent set of correction elements under the same configuration version. Based on the correction configuration group, response mapping, dark-field restoration, and scattering baseline correction are performed on the original image to generate a corrected image. The corrected image is then associated with a set of frame numbers. Specifically, each original image in the frame number set undergoes the correction process sequentially. First, response mapping is performed: the grayscale of the original image is mapped pixel-by-pixel to the linear response domain according to the response mapping table, eliminating grayscale compression or stretching caused by detector nonlinearity. Next, dark-field restoration is performed: the dark field is subtracted pixel-by-pixel from the mapped image to remove background bias, and then the pixel gain is normalized according to the difference between the flat and dark fields, completing pixel-level gain uniformity processing. Finally, scattering baseline correction is performed: background compensation is applied to the image based on the scattering field baseline, suppressing the gradually varying background caused by fixed scattering while preserving local contrast caused by structural changes. After each frame is processed, a corresponding corrected image file is generated, and the corrected image is associated with its frame number and written into the correction result table. The correction result table also records the compensation number reference, the frame sequence consistency mark reference, and the original image reference, forming a stable link of "frame sequence number - frame number - original image - corrected image - compensation version", which provides a consistent input basis for subsequent bad pixel replacement, artifact separation, standardized image generation, and structural index generation.

[0028] The present invention is further configured such that the artifact suppression and structure indexing include: The preprocessing correction module generates a bad pixel replacement image based on the bad pixel mask of the corrected image, and establishes a relationship between the bad pixel replacement image and the frame number set. Specifically, it reads the correction result table to obtain the frame number set and the corresponding corrected image, and reads the bad pixel mask from the correction configuration group. For each frame of corrected image, the bad pixel mask marking is checked pixel by pixel: the bad pixel location is not directly involved in subsequent analysis, but is instead filled with a neighborhood replacement value. The generation of the replacement value adopts a "restricted neighborhood reconstruction" process: first, a neighborhood window of a preset shape is selected around the bad pixel, and pixels within the window that are also marked as bad pixels are removed; then, the remaining valid pixels are weighted according to their distance from the bad pixel, with closer pixels having higher weights; at the same time, consistency constraints are applied to the edge direction to avoid filling in pixels across obvious gray-level abrupt boundaries; finally, the weighted and consistent valid pixels are used to generate bad pixel replacement values ​​and backfill them into the bad pixel location. For areas with continuous clusters of bad pixels, a "progressive filling from the outside in" order is adopted, filling in the boundary bad pixels first and then the internal bad pixels to prevent the internal areas from lacking effective references. After each frame is processed, a bad pixel replacement image is generated, and the bad pixel replacement image is associated with the corresponding frame number and written into the replacement result table. The replacement result table also retains the index of bad pixel mask version and corrected image reference. Based on the defective pixel replacement image, a fixed pattern term and a structural response term are generated to form an artifact separation image. The artifact separation image is then associated with the compensation number. Specifically, the defective pixel replacement image is obtained by reading the replacement result table, and the compensation number references the corresponding compensation configuration version information. Artifact separation is performed on each frame of the defective pixel replacement image: First, the fixed pattern term is estimated to characterize stable artifacts unrelated to structure, such as detector stripes, rings, and gradually varying brightness drift. The estimation process employs a "strong smoothing constraint and anomaly suppression" strategy, requiring the fixed pattern term to change slowly in space and penalizing local abrupt changes to avoid erroneously absorbing structural edges into the fixed pattern term. Then, the fixed pattern term is subtracted from the defective pixel replacement image to obtain the structural response term, which emphasizes local contrast and edge information related to the internal structure of the clamp. Subsequently, a uniformization process is performed on the structural response term to limit the diffusion of residual fixed pattern artifacts between different frames, ensuring the artifact separation result remains stable across frames. The fixed mode item reference, structural response item reference, and the resulting artifact separation image are written into the artifact separation table, and the artifact separation table is associated with the compensation number to ensure that the artifact processing result can be traced back to the specific compensation version and bad pixel replacement process. Based on artifact separation images, standardized images and quality labels are generated, and these standardized images and quality labels are associated with frame sequence numbers. Specifically, the artifact separation table is read to obtain the artifact-separated images, and the frame sequence number and frame number set reference are read. Standardization generation is performed on each frame of artifact-separated images: first, grayscale dynamic range remapping is performed to compress pixel values ​​to a uniform range. The compression strategy adopts the rule of "suppressing extreme values ​​and retaining intermediate contrast" to avoid a small number of bright or dark pixels dominating the global scale; then, local contrast correction is performed to ensure that structural details under different viewing angles and exposure conditions are at a comparable scale; finally, the standardized image is output and the standardized configuration reference is recorded. Quality labels are generated simultaneously, and the quality labels consist of several quality elements: first, the sharpness element, which represents the degree of blur by statistically analyzing the concentration of edge responses; second, the noise suppression element, which represents the level of artifact persistence by statistically analyzing the residual fluctuations in unstructured areas; and third, the coverage integrity element, which represents the occlusion or saturation effect by statistically analyzing the proportion of available pixels in the effective detection domain. Based on the dereference of extrinsic parameter numbers, the projection coefficient set is obtained. A structural index is generated for the standardized image. The standardized image, structural index, quality identifier, compensation number, extrinsic parameter number, and frame sequence number are written into the preprocessing record, and a preprocessing consistency mark is generated. Specifically, the extrinsic parameter number in the acquisition record is read, and the projection coefficient set and its version information are obtained by dereferencing it in the extrinsic parameter version library. Structural index generation is performed on the standardized image: first, the image coordinates are mapped to structural coordinates according to the projection coefficients, so that the same structural part under different views is aligned in the structural coordinate domain; then, the boundary definition of the clamp structure section is read according to the structural template library, and the pixels whose structural coordinates fall into different sections are assigned section index numbers respectively, forming a structural index map. The structural template library is indexed according to the clamp model and tooling configuration to ensure that the structural section division corresponds to the actual hardware structure. Subsequently, a preprocessing consistency mark is generated: the preprocessing consistency mark is bound and encoded according to five elements: "compensation version, extrinsic parameter version, frame sequence organization, standardization result, and structural index result", so that any change in any element can be reflected in the consistency mark.

[0029] The present invention is further configured such that the defect determination module includes: The defect determination module reads the preprocessing record to obtain standardized images, structural indexes, and frame numbers, generates a frame alignment identifier, and establishes a link with the preprocessing record. Specifically, upon entering the defect determination stage, it reads standardized image references, structural index references, and frame number list references from the preprocessing record, and dereferences them in frame number order to obtain paired "standardized image-structural index" frame sets. A frame alignment identifier is generated for this frame set, following the rules of "order sensitivity, content sensitivity, and verifiability": first, the frame number, standardized image identifier, and structural index identifier of each frame are concatenated in a fixed order to form a frame-level summary; then, all frame-level summaries are concatenated in frame number order to form a sequence summary; finally, the sequence summary is compressed and encoded to obtain the frame alignment identifier. During the generation process, a consistency check is performed simultaneously to check whether the same frame number is duplicated, whether there are missing numbers, and whether there are version inconsistencies in the standardized image and structural index references. The check results are written to an auxiliary field of the frame alignment identifier. The frame alignment identifier is written to the alignment record and linked with the preprocessing record, serving as a unified primary key for subsequent detection domains, saliency maps, and candidate domains. A detection domain indicator is generated based on the structural index and structural segment set identifiers, and then associated with the corresponding frame number. Specifically, the segment list corresponding to the structural segment set identifier is read. This list is provided by the structural template library and specifies the range of structural segments participating in defect judgment, such as compression areas, wedge areas, and transition areas. For each frame's structural index, the segment number is checked pixel by pixel to see if it belongs to the segment list. If it does, it is marked as a valid detection position; otherwise, it is marked as a non-detection position, thus forming the detection domain indicator. To avoid detection domain breaks caused by boundary jitter, boundary consistency processing is performed on the detection domain indicator: first, connectivity checks are performed near the segment boundaries to remove isolated small segments; then, narrow break zones are bridged to keep the detection domain continuous at the structural boundaries; at the same time, the boundary lines between structural segments are preserved to prevent cross-segment expansion. The detection domain indicator of each frame is associated with the corresponding frame number and written into the detection domain record. The detection domain record is also associated with the frame alignment identifier to ensure that subsequent saliency map generation is carried out under the same detection domain constraint. A structural constraint saliency map is generated based on standardized images and detection domain indicators, and then associated with the corresponding frame number. Specifically, each frame of standardized imagery and its detection domain indicator are read, and saliency responses are generated for pixels within the detection domain. The saliency responses are constructed using a logic of "structural boundary sensitivity, gradual background suppression, and local anomaly prominence": First, a detail response map is generated on the standardized imagery. The detail response is obtained through multi-scale edge enhancement and curvature enhancement, making structural abrupt changes such as crack boundaries, pore contours, and interface debonding edges more prominent. Then, a background suppression map is generated. The background suppression map reduces the contribution of scattering residues and fixed pattern residues to saliency by stripping away gradual variation components. Subsequently, the detail response and background suppression are constrained and fused, retaining saliency responses only at locations where the detection domain indicator is marked as valid, and suppressing saliency responses to low values ​​at non-detection locations, thus forming the structural constraint saliency map. To ensure cross-frame stability, inter-frame consistency constraints are applied to the structural constraint saliency map: consistency checks are performed on the saliency responses of the same structural segment in adjacent frames, weakening isolated spikes that only appear in a single frame and retaining stable saliency regions that recur in multiple frames. Candidate regions and their location indices are generated based on the structural constraint saliency map. These candidate regions and their location indices are then associated with the structural index and written into the candidate records. Specifically, the structural constraint saliency map in the saliency records is read, and candidate region extraction is performed on salient responses within the detection domain. Candidate region extraction employs a combined area constraint and connectivity constraint process: first, the target candidate area range is determined based on the allowable candidate size of the structural segment; then, a threshold satisfying the target area range is selected on the saliency map, ensuring that the total area of ​​connected regions above the threshold falls within a preset range; morphological constraint filtering is performed on the connected regions obtained from the threshold segmentation, removing regions with excessively fine shapes and regular stripes to reduce residual artifacts, and removing small, isolated noise regions to reduce false detections; regions satisfying both connectivity and morphological constraints are retained as candidate regions.

[0030] The present invention is further configured such that the segment evidence discrimination includes: The defect determination module reads candidate records to obtain candidate domain location indices, generates structural segment-level evidence based on the structural constraint saliency map using the structural index, and establishes a correlation between the structural segment-level evidence and frame alignment identifiers. Specifically, it reads candidate records to obtain candidate domain location indices, simultaneously reads saliency records to obtain structural constraint saliency maps, and reads preprocessed records to obtain structural indices and frame alignment identifiers. Based on the candidate domain location indices, it locates the corresponding pixel range of the candidate domain in each frame, and then, based on the structural index, it groups the pixels within the candidate domain according to the structural segment number, forming a "segment-level candidate pixel set." For each structural segment, it calculates the structural segment-level evidence, which is formed using a "combined intensity accumulation and morphological penalty" process: first, it accumulates the saliency responses of candidate pixels within the segment, assigning higher contributions to pixels with strong and continuous saliency responses; then, it penalizes the morphology of the candidate domain. If the candidate domain exhibits regular stripes or long, straight, narrow band features, the evidence weight of that segment is reduced; if the candidate domain exhibits irregular shapes such as closed boundaries, local voids, or crack forks, the effective weight of the evidence weight is increased. The obtained structural segment-level evidence quantity is written into the segment evidence record, and the segment evidence record is associated with the frame alignment identifier to ensure that the evidence quantity can be traced back to the specific frame set and the corresponding preprocessing record. A cross-frame consistency factor is generated based on the amount of evidence at the structural segment level, and this cross-frame consistency factor is associated with the structural segment set identifier. Specifically, segment evidence records are read, and the evidence quantity sequences of the same structural segment in different frames are merged according to the structural segment set identifier. A cross-frame consistency factor is generated for the evidence quantity sequence, following the rule of "enhancing repetition, suppressing isolated occurrence, and penalizing amplitude drift": first, the proportion of frames in which the same structural segment has valid evidence in the frame sequence is counted; the more times it appears, the higher the consistency. Then, the amplitude of the change in evidence quantity between adjacent frames is checked; if the change is too drastic, the consistency is reduced. At the same time, it is checked whether the relative position of the evidence peak position within the structural segment remains stable with frame changes; if the position drift is significant, the consistency is further reduced. The cross-frame consistency factor is written into the consistency record and associated with the structural segment set identifier, so that subsequent defect type determination can reference the consistency result at the segment level. Defect type identifiers are generated based on structural segment-level evidence quantity, cross-frame consistency factors, and defect type fingerprint parameters. These defect type identifiers are then associated with structural segment set identifiers. Specifically, cross-frame consistency factors are obtained by reading consistency records, structural segment-level evidence quantity is obtained by reading segment evidence records, and defect type fingerprint parameters are retrieved from the defect type fingerprint database. The defect type fingerprint parameters, organized by structural segments, describe the evidence distribution pattern and allowable fluctuation range of different defect types across different segments. For each defect type, a matching and discrimination process is performed: the structural segment-level evidence quantity is compared segment by segment with the fingerprint parameters for that type, focusing on whether the evidence quantity meets the characteristic requirements of that type in key segments and whether it is within the allowable range in non-key segments. The cross-frame consistency factor is then used as a credibility constraint; only when the consistency threshold is met is the evidence from that segment allowed to enter the valid set of type matching. All matching results are sorted, and the type with the highest ranking and satisfying the mutual exclusion constraint is selected as the defect type identifier. If multiple types have similar scores, a decision is made according to the priority rules specified by the structural segment set identifiers to avoid type stacking. Write the defect type identifier into the type record and associate it with the structural segment set identifier; Evidence references and location index identifiers are generated based on the session number, frame alignment identifier, candidate domain location index, and defect type identifier. The defect type identifier, location index identifier, and evidence reference are written into the judgment record and associated with the preprocessing record. Specifically, the session number is obtained by reading the session record, the frame alignment identifier is obtained by reading the alignment record, the candidate domain location index is obtained by reading the candidate record, and the defect type identifier is obtained by reading the type record. The process of generating evidence references and location index identifiers adopts the "link element binding encoding" rule: the session number is used as the session-level primary key, the frame alignment identifier as the frame set-level primary key, the candidate domain location index as the spatial positioning element, and the defect type identifier as the semantic positioning element, combined in a preset order to form the evidence reference. Simultaneously, the candidate domain location index is re-encoded at the structural segment level to form the location index identifier, enabling the location index identifier to point to the specific structural segment and candidate domain boundary range. The defect type identifier, location index identifier, and evidence reference are written into the judgment record, and the preprocessing record reference identifier and frame alignment identifier reference identifier are also written into the judgment record, ensuring that the judgment result can be traced back to the original evidence along the "judgment record—preprocessing record—standardized image—structural index—frame sequence" link.

[0031] The present invention is further configured such that the defect determination module includes: Based on the judgment record, defect type identifier, location index identifier, and evidence reference identifier are obtained, forming an archived field set and associating it with the session number. Specifically, during the archive tracing stage, the defect type identifier, location index identifier, and evidence reference identifier are read from the judgment record, along with the preprocessing record reference and session number reference registered in the judgment record. An integrity check is performed on the three types of identifiers read, checking whether the identifier is empty, whether there are duplicate entries, and whether the evidence reference identifier can be traced back to the corresponding candidate domain location index. After passing the check, the defect type identifier, location index identifier, and evidence reference identifier are assembled into an archived field set entry according to the same entry organization rules. The archived field set contains the entry sequence number, the three types of identifier values, and their source record references. A primary key association is established between the archived field set and the session number, and written into the archived field table, so that subsequent session association keys, acquisition association keys, and preprocessing association keys can all index the archived field set through the session number. A session association key is generated based on the session number, status code, and view sequence number. An acquisition association key is generated based on the session number, external parameter number, exposure number, and frame sequence number. A preprocessing association key is generated based on the session number, compensation number, standardized image identifier, and structure index identifier. These three types of association keys are associated with archived field sets. Specifically, the status code and view sequence number are obtained by reading session records; the external parameter number, exposure number, and frame sequence number are obtained by reading acquisition records; and the compensation number, standardized image identifier, and structure index identifier are obtained by reading preprocessing records. The three types of association keys are generated using a "fixed-order combination, intra-domain compression, and conflict resolution" logic: first, the participating elements are arranged in a preset order to form combined segments; then, the combined segments are compressed into fixed-length key values, and conflict checks are performed. If a key value conflict occurs, a version count is appended as a resolution field to ensure that the key value is unique within the same session domain. The session association key is generated by combining the session number, status code, and view sequence number; the acquisition association key is generated by combining the session number, external parameter number, exposure number, and frame sequence number; and the preprocessing association key is generated by combining the session number, compensation number, standardized image identifier, and structure index identifier. Write the three types of association keys into the association key record, and establish a reference relationship between the association key record and the archived field set with the session number as the bridge, so that the archived field set entries can simultaneously point to the corresponding versions of the session link, the acquisition link and the preprocessing link; An evidence mapping index is generated based on the evidence citation identifier, location index identifier, defect type identifier, and frame sequence number. This index is then associated with standardized image identifiers and structural index identifiers. Specifically, the evidence citation identifier, location index identifier, and defect type identifier are read from the archived field set, the frame sequence number is read from the acquisition record, and the standardized image identifier and structural index identifier are read from the preprocessing record. An evidence mapping index is generated for each archived field set entry, using an "evidence element binding" logic: the evidence citation identifier is used as the primary key, the location index identifier and defect type identifier as semantic and spatial limiting elements, and the frame sequence number as the frame set limiting element, forming a verifiable mapping entry. The mapping entry also includes the standardized image identifier and structural index identifier as evidence location references, allowing the evidence citation identifier to directly locate the specific standardized image version and structural index version through the evidence mapping index. The evidence mapping index is written into the evidence mapping table and a one-to-one reference relationship is established with the archived field set entries, ensuring that subsequent tracing relationships can be constructed by restoring the evidence chain along the evidence mapping table. A set of traceability edges is generated based on the archived field set, and a graph signature is generated, which is associated with the archived field set. Specifically, the archived field set, associated key records, and evidence mapping table are read, and a set of traceability edges is generated according to a preset relationship template. The construction of the set of traceability edges follows the directed relationship rule of "primary key to version, version to evidence, and evidence to result": the session number points to the session association key, acquisition association key, and preprocessing association key; the acquisition association key points to the external parameter number, exposure number, and frame sequence number; the preprocessing association key points to the compensation number, standardized image identifier, and structure index identifier; the evidence reference identifier points to the evidence mapping index, location index identifier, and defect type identifier; at the same time, a frame-to-image reference edge is established between the frame sequence number and the standardized image identifier, and an external parameter-to-structure reference edge is established between the external parameter number and the structure index identifier. After completing the edge set, a graph signature is generated. The graph signature generation process employs "ordered edge encoding and full compression": first, edge encoding fragments are generated for each edge according to the encoding rules for the start and end points; then, these edge encoding fragments are sorted according to a unified sorting rule; finally, the sorted edge encoding fragments are compressed as a whole to form the graph signature, ensuring that any addition, deletion, or modification to any edge will result in a change to the graph signature. The graph signature is written to a graph signature record, and the graph signature record is associated with an archived field set and a session number, ensuring that the archived field set has a verifiable traceable structure digest before the archived record is written.

[0032] The present invention is further configured such that the archive generation index includes: The archiving and tracing module generates an archive list based on the session association key, acquisition association key, preprocessing association key, image signature, and evidence mapping index. It then generates an archive number based on the archive list and associates it with the session number. Specifically, the archiving and tracing module reads the session association key, acquisition association key, preprocessing association key, image signature, and evidence mapping index, while also reading the session number and status code reference from the session records. These elements are assembled into an archive list in a fixed order. The archive list consists of two parts: a header and entries. The header records the session number, status code, session association key, acquisition association key, preprocessing association key, and image signature. Each entry lists the evidence reference identifier, location index identifier, defect type identifier, frame sequence number reference, and corresponding standardized image identifier and structure index identifier reference, according to the evidence mapping index. A partition index is generated based on the device identifier and time count. A path index is then generated based on the partition index, and the partition index and path index are associated with the archive number. Specifically, the device identifier is obtained by reading the device registration information, the time count is obtained by reading the system clock, and the archive number is obtained by reading the archive number record. The partition index is generated using a "device-dimensional bucketing, time-dimensional slicing" rule: first, the device identifier is mapped to a preset bucket range to form a device bucket number; then, the time count is sliced ​​by day or shift to form a time slice number. The two are combined to obtain the partition index, ensuring that archive records of the same device within the same time slice fall into the same partition. The path index is generated using a "sequential location within partitions and conflict resolution" rule: the path index is formed by adding a summary fragment of the archive number as a prefix to the partition index; if a collision occurs in the path index, an incrementing sequence number is added as a resolution field until the path index is unique. The partition index and path index are written to the index record and associated with the archive number. The index record also stores a reference to the target storage path for easy location during subsequent archive record writing. An archive consistency marker is generated based on the archive number, session number, status code, path index, and evidence citation identifier, and then associated with the archive number. Specifically, the archive number, session number, status code, path index, and the evidence citation identifier set from the archive list entries are read. The generation of the archive consistency marker adopts a "link element binding" logic: the archive number is used as the primary key element, the session number and status code as session constraint elements, and the path index as the storage location element. Simultaneously, a set summary of the evidence citation identifier set is generated according to a fixed sorting rule as the evidence constraint element. These elements are combined in a fixed order and compressed to obtain the archive consistency marker. Consistency verification is performed during the generation process, checking whether the archive consistency marker can be reverse-located to the same archive number and whether the evidence citation identifier set summary is consistent with the archive list entries. The archive consistency marker is written to a consistency marker record and associated with the archive number to ensure that subsequent retrieval or migration can verify that the archived content has not been replaced or mismatched. Archived records are generated based on the archive number, session number, and archive consistency marker, and an association is established between the session number and the status code index. Specifically, the archive number record, index record, and consistency marker record are read to assemble the archived record. The archived record contains the archive number, session number, status code, session association key, acquisition association key, preprocessing association key, image signature, evidence mapping index reference, partition index, path index, and archive consistency marker. It also includes references to session records, acquisition records, preprocessing records, judgment records, and the archived list. After writing the archived record to the archive database, a session number index and a status code index are established: the session number index is used to retrieve archived records by session dimension, and the status code index is used to group and filter archived records by permission or non-permission status. The index entries store the archive number and path index to ensure that the storage location of the archived record can be directly located after a search hit. At this point, the entire closed-loop process of generating the archived index is completed, and the archived records have verifiable associations in the session domain, evidence domain, and storage domain.

[0033] It should be noted that the specific operation methods of each module and unit in the X-ray digital imaging automatic inspection system for tension clamps of multi-split transmission lines provided in the above embodiments have been described in detail in the system embodiments and will not be repeated here. In practical applications, the X-ray digital imaging automatic inspection system for tension clamps of multi-split transmission lines provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above, and this is not a limitation.

[0034] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors, characterized in that, include: The gating planning module collects radiation status and stability to obtain status codes, and generates session numbers, view sequence numbers and session records. The calibration acquisition module reads the session record. If the status code is in the permitted state, it generates the external parameter number, exposure number, view sequence number, acquisition frame sequence number, and generates the acquisition record. The preprocessing and correction module reads the acquisition records to obtain the compensation number, external parameter number, and frame sequence number, and generates standardized images, structural indexes, and preprocessing records. The defect determination module obtains structural indexes and standardized images based on preprocessed records, generates defect types and evidence references, and writes them into the determination record. The archive traceability module reads session records, collection records, preprocessing records, and judgment records, and writes them into the archive records along with the session number and status code.

2. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines according to claim 1, characterized in that, The gating planning module includes: The dose rate sequence, interlocking markers, and source preparation markers are collected to form a radiation state record, and the displacement sequence and acceleration amplitude sequence are collected to form a stability record. Based on the upper limit of the measurement range, normalized sequences are generated from the dose rate sequence, displacement sequence, and acceleration amplitude sequence and written into the gated intermediate record; A radiation compliance index is generated based on the normalized sequence and dose limit; a stability index is generated based on the normalized sequence and displacement and vibration limits; and a gating discrimination quantity is generated based on the radiation compliance index and the stability index. Status codes are generated from gating discriminants based on radiation threshold, stability threshold, and discrimination threshold. A session number is generated based on the task identifier, device identifier, time count, and status code. A view sequence number is generated based on the view index sequence. The session number, view sequence number, and status code are written into the session record.

3. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors according to claim 2, characterized in that, The calibration acquisition module includes: The calibration acquisition module reads the session record to obtain the session number, view sequence number, and status code; When the status code is in the permitted state, a view index sequence is formed based on the view sequence number, and an association identifier is established between the view index sequence and the session number; Based on the viewpoint index sequence, set the calibration posture and collect calibration frames, then establish an association between the calibration frames and the viewpoint index; Based on the calibration frame, target feature coordinates and target feature confidence values ​​are extracted, and the target feature coordinates and target feature confidence values ​​are associated with the session number and view index.

4. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines according to claim 3, characterized in that, External parameter exposure frame sequence records include: The calibration acquisition module generates a projection coefficient set based on the target spatial coordinates, target feature coordinates and target feature confidence, generates an external parameter number based on the projection coefficient set, and writes the external parameter number, session number and view sequence number into the acquisition record. An exposure number is generated based on the set of exposure levels and the upper limit of the dose. The exposure number, session number, and external parameter number are written into the acquisition record. Frame sequence numbers are generated by collecting frame numbers based on the viewpoint index sequence, and the frame sequence number, viewpoint sequence number, and exposure number are written into the acquisition record. A consistency marker is generated based on the session number, external parameter number, exposure number, frame sequence number, and status code. The consistency marker is written into the acquisition record and associated with the status code.

5. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines according to claim 1, characterized in that, The preprocessing correction module includes: The preprocessing correction module reads the acquisition record to obtain the compensation number, external parameter number, and frame sequence number, dereferences the frame sequence number to obtain the frame number set and the corresponding original image, and establishes a relationship between the frame number set and the frame sequence number. A frame sequence consistency marker is generated based on the frame number set, and the frame sequence consistency marker is associated with the compensation number, external parameter number, and frame sequence number. Based on the dereference of the compensation number, the dark field, flat field, bad pixel mask, scattering field baseline, and response mapping table are obtained to form a correction configuration group, and the correction configuration group is associated with the compensation number. Based on the correction configuration group, the original image is subjected to response mapping, flat dark field restoration and scattering baseline correction to generate a corrected image, and the corrected image is associated with the frame number set.

6. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors according to claim 5, characterized in that, Artifact suppression and structure indexing include: The preprocessing correction module generates a bad pixel replacement image for the corrected image based on the bad pixel mask, and establishes a relationship between the bad pixel replacement image and the frame number set. Based on the bad pixel replacement image, a fixed pattern term and a structural response term are generated to form an artifact separation image, and the artifact separation image is associated with the compensation number. Based on artifact separation images, standardized images and quality labels are generated, and the standardized images and quality labels are associated with frame sequence numbers; The projection coefficient set is obtained by dereferencing the extrinsic parameter number. A structural index is generated for the standardized image. The standardized image, structural index, quality identifier, compensation number, extrinsic parameter number, and frame sequence number are written into the preprocessing record and a preprocessing consistency mark is generated.

7. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors according to claim 1, characterized in that, The defect determination module includes: The defect determination module reads the preprocessing record to obtain standardized images, structural indexes, and frame numbers, generates frame alignment identifiers, and establishes a correlation with the preprocessing record. A detection domain indicator is generated based on the structure index and the structure segment set identifier, and the detection domain indicator is associated with the corresponding frame number. A structural constraint saliency map is generated based on standardized images and detection domain indicators, and the structural constraint saliency map is associated with the corresponding frame number. Candidate domains and candidate domain location indices are generated based on the structural constraint saliency map, and the candidate domains, candidate domain location indices, and structural indexes are associated and written into candidate records.

8. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors according to claim 7, characterized in that, Section evidence identification includes: The defect determination module reads candidate records to obtain candidate domain position indexes, generates structural segment-level evidence based on the structural constraint saliency map according to the structural index, and establishes a correlation between the structural segment-level evidence and the frame alignment identifier. Based on the amount of evidence at the structural segment level, a cross-frame consistency factor is generated, and the cross-frame consistency factor is associated with the structural segment set identifier. Defect type identifiers are generated based on structural segment-level evidence quantity, cross-frame consistency factor, and defect type fingerprint parameters, and the defect type identifiers are associated with structural segment set identifiers. Evidence references and location indexes are generated based on session number, frame alignment identifier, candidate domain location index, and defect type identifier. The defect type identifier, location index identifier, and evidence reference are written into the judgment record and associated with the preprocessing record.

9. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission conductors according to claim 1, characterized in that, The defect determination module includes: Based on the judgment record, the defect type identifier, location index identifier, and evidence citation identifier are obtained to form an archived field set and associate it with the session number; A session association key is generated based on the session number, status code, and view sequence number; an acquisition association key is generated based on the session number, external parameter number, exposure number, and frame sequence number; and a preprocessing association key is generated based on the session number, compensation number, standardized image identifier, and structure index identifier. The three types of association keys are associated with the archive field set. An evidence mapping index is generated based on the evidence citation identifier, location index identifier, defect type identifier, and frame sequence number. The evidence mapping index is then associated with the standardized image identifier and the structural index identifier. Generate a set of traceability edges based on the archived field set, generate a graph signature, and associate the graph signature with the archived field set.

10. The automatic X-ray digital imaging inspection system for tension clamps of multi-split transmission lines according to claim 1, characterized in that, The archived index includes: The archiving and tracing module generates an archive list based on the session association key, the collection association key, the preprocessing association key, and the image signature and evidence mapping index. It then generates an archive number based on the archive list and associates it with the session number. A partition index is generated based on the device identifier and time count. A path index is generated based on the partition index. The partition index and the path index are associated with the archive number. Generate an archive consistency tag based on the archive number, session number, status code, path index, and evidence citation identifier, and associate the archive consistency tag with the archive number; Archive records are generated based on archive number, session number, and archive consistency marker, and an association is established between session number and status code index.

Images