Method for improving detection precision of lead sealing defect part of cable based on full-matrix capture and full-focus algorithm

Abstract

The invention discloses a method for improving the detection precision of a cable lead seal defect part based on a full-matrix capture and full-focus algorithm, which comprises the following steps of: scanning a cable lead seal by using a two-dimensional phased array probe to obtain full-matrix capture FMC data sets, processing each obtained FMC data set by using a single-medium full-focus TFM algorithm to obtain a full-matrix capture FMC data set; the initial surface contour and the final surface contour of the target object are obtained, and then surface verification is carried out to obtain the final three-dimensional surface contour of the target object; then calculating new flight time, determining a ray path by using the Fermat shortest time principle, calculating image intensity of internal points so as to obtain a complete internal image of the target image, then carrying out image splicing, splicing all internal TFM images together, creating a three-dimensional TFM image in the whole sample, and finally obtaining a three-dimensional TFM image in the whole sample. Therefore, the final three-dimensional surface contour of the complete target object and the reconstruction of the internal image are obtained, and the detection precision of the internal defect of the cable lead seal is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of nondestructive testing, and in particular relates to a method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm. Background Art

[0002] There are many limitations to NDT inspections using traditional single-element probes, including the inability to adjust the depth of focus, limited directions in which defects can be detected, and the resulting potential for missing defects in unexpected locations or orientations. Phased arrays can overcome these issues. Ultrasonic phased arrays consist of multiple single-element transducers and have the advantages of electronic beam deflection, focusing, and scanning, which are achieved by applying delay laws to individually addressable elements; due to these characteristics, phased arrays can be applied to a wider range of imaging scenarios, significantly speed up ultrasonic inspections, and improve defect image resolution. However, current phased array-based NDT inspections are mainly limited to one-dimensional (1D) linear element arrays, so only two-dimensional (2D) slices of a component can be imaged from a single array position. An important limitation of 1D arrays is that steering is limited to one plane, so they cannot accurately focus through multi-directionally curved surfaces. There is also the potential to miss scattering effects in the unreconstructed dimensions, which means there is a potential for missing defects in unexpected locations or orientations.

[0003] Creating a 2D array containing elements distributed across a two-dimensional aperture allows for better characterization of defects within a component. Similar to how a 1D array allows for beam steering within a 2D plane, a 2D array can manipulate the beam within a 3D volume without moving the array. This capability can significantly increase coverage of the imaging area and allow for more detailed inspections through voxel-based beam steering and focusing, as long as the array and surface orientation are favorable. Distinguishing between defect types is critical for a thorough and accurate inspection because the severity of a defect depends on its size, orientation, and shape; for example, planar discontinuities such as cracks are known to be generally more dangerous than volumetric defects such as voids due to their sharp edges that may grow and lead to fracture. Therefore, obtaining a 3D image of the defect is important in order to accurately characterize the defect.

[0004] The state of the art is to inspect for defects in areas where the surface geometry is curved in multiple directions (i.e., hyperboloids), such as those found in pipe branches or nozzles. Current inspection procedures through these surfaces involve the use of either a single element transducer that probes the area from a series of positions, or a radiometric method. The use of a single element transducer means that an experienced operator is required to interpret the data, and constructing a volumetric image of the area is extremely difficult; while radiometric methods are not effective for detecting and sizing planar defects without prior knowledge of their likely location and orientation. Due to the presence of hyperboloids, 1D arrays are not suitable for the inspection of such surfaces. Therefore, the ability of 2D arrays to focus through hyperboloids provides a huge opportunity for improvement in defect detection in complex shaped components. Therefore, it is necessary to design a method based on full matrix capture and full focusing algorithm to improve the detection accuracy of cable seal defect locations to solve the above problems. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method for improving the detection accuracy of cable seal defect parts based on full matrix capture and full focusing algorithm, aiming to solve the problem of low detection accuracy of cable seal defects identified by ultrasonic phased array in the prior art.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm includes: S1, the cable seal is scanned by a two-dimensional phased array probe to obtain a full matrix capture FMC data set; S2, using the single-medium total focusing TFM algorithm to process each obtained FMC data set to obtain the preliminary surface profile and the final surface profile of the target object; performing surface verification to obtain the final three-dimensional surface profile of the target object; S3, calculate the new flight time TOF, determine the ray path based on the Fermat shortest time principle, calculate the image intensity of the internal points, and obtain the complete internal imaging of the target image; S4, image stitching is performed to stitch all internal TFM images together to create a three-dimensional TFM image of the entire interior of the sample, and ultimately obtain the final three-dimensional surface contour and internal image reconstruction of the complete target object.

[0007] Preferably, in step S1, scanning the cable seal to obtain a full matrix capture FMC data set includes: A 2D sparse array of 128 elements was constructed. The elements were arranged in a Poisson disk distribution to avoid grating effects and maintain high resolution. The operating frequency of the array was set to 3 MHz. Each element was a circle with a diameter of 1.7 mm. The spacing between adjacent elements was 0.2 mm. The aperture diameter of the entire array was 30 mm. Data collection was performed at each array position for 10 average values ​​to ensure data stability and accuracy; Use a Gaussian window function centered on the array center frequency for filtering and Hilbert transform to process the time domain signal, with a 40dB half bandwidth of 90% of the array center frequency; The time domain signal of each transmit-receive element pair is captured and stored to form an FMC data set.

[0008] Preferably, in step S2, processing each FMC data set using a single-medium total focusing TFM algorithm comprises: Setting a size threshold of an imaging grid and creating an imaging grid whose size is larger than the set threshold; Calculate the intensity of each image point P1 ; Determine the location of the global maximum amplitude as the first point of the surface; Perform two passes, the first pass with the main direction being x and the secondary direction being y; the second pass with the main direction being y and the secondary direction being x. Finally, average the two passes to obtain two surface contour estimates. and .

[0009] Preferably, performing surface verification in step S2 to obtain the final three-dimensional surface profile of the target object includes: The final surface profile to be extracted Compared with the real surface generated by the Gaussian function equation, the z position error of each surface point is obtained to obtain the final extracted 3D surface profile.

[0010] Preferably, in step S3, calculating a new time of flight TOF, determining the ray path based on the Fermat shortest time principle, calculating the image intensity of the internal points, and obtaining a complete internal imaging of the target image includes: For each array position, a subset of extracted surface points is considered to calculate the TOF; The Fermat shortest time principle is used to determine the two surface intersection points that produce the global minimum total TOF; Calculate the distance from the extracted surface point to any image point TOF and store the TOF from each surface point into an array element TOF; Determine from arrive The true projection path represents the actual ultrasonic path, thereby obtaining a new TOF set.

[0011] Preferably, in step S3, calculating the image intensity of the internal point includes: This is accomplished by summing the intensity of all transmitting and receiving elements to calculate the intensity of each internal image point in the imaging grid. ; thus obtaining a 3D TFM internal image of each array position.

[0012] Preferably, in step S4, stitching all the internal TFM images together to create a three-dimensional TFM image of the entire interior of the sample includes: Input 3D TFM interior images from multiple array positions and create a 3D TFM image of the entire sample interior by stitching all interior TFM images together, resulting in a complete 3D TFM interior image.

[0013] Preferably, after obtaining the final three-dimensional surface contour and internal image reconstruction of the complete target object in step S4, the obtained results are analyzed, specifically including: The final extracted 3D surface profile imaging is evaluated, the z position error between the extracted surface and the real surface is compared, and the RMS error and other statistical indicators are calculated. The internal imaging results are evaluated, the position of visible defects is measured, their depth and lateral distance are compared, and the imaging accuracy is evaluated. The stitched 3D surface TFM image is displayed as an isosurface with color indicating depth, and the stitched 3D internal TFM image is displayed as an isosurface with color indicating distance.

[0014] The beneficial effects of the present invention are as follows: This method uses a two-dimensional phased array probe to scan the cable seal to obtain a full matrix capture FMC data set, and uses a single-medium full-focus TFM algorithm to process each FMC data set to obtain the preliminary surface profile and final surface profile of the target object, and then performs surface verification to obtain the final three-dimensional surface profile of the target object; then calculate the new flight time TOF, use the Fermat shortest time principle to determine the ray path, calculate the image intensity of the internal point, and obtain the complete internal imaging of the target image, and then perform image stitching to stitch all the internal TFM images together to create a three-dimensional TFM image of the entire sample. Thus, the final three-dimensional surface profile and internal image of the complete target object are reconstructed, which effectively improves the detection accuracy of internal defects of cable seals and solves the problem of low detection accuracy of ultrasonic phased array identification of cable seal defects in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 A flow chart of a control method for extracting preliminary and final surface contours from an FMC data set provided by an embodiment of the present invention; Figure 3A flowchart of a method for controlling an extraction process from extracted surface points and FMC data sets to internal 3D TFM image generation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] Embodiment 1: like Figure 1 As shown, the method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm includes: S1, the cable seal is scanned by a two-dimensional phased array probe to obtain a full matrix capture FMC data set; S2, using the single-medium total focusing TFM algorithm to process each obtained FMC data set to obtain the preliminary surface profile and the final surface profile of the target object; performing surface verification to obtain the final three-dimensional surface profile of the target object; S3, calculate the new flight time TOF, determine the ray path based on the Fermat shortest time principle, calculate the image intensity of the internal points, and obtain the complete internal imaging of the target image; S4, image stitching is performed to stitch all internal TFM images together to create a three-dimensional TFM image of the entire interior of the sample, and ultimately obtain the final three-dimensional surface contour and internal image reconstruction of the complete target object.

[0017] Preferably, in step S1, scanning the cable seal to obtain a full matrix capture FMC data set includes: A 2D sparse array of 128 elements was constructed. The elements were arranged in a Poisson disk distribution to avoid grating effects and maintain high resolution. The operating frequency of the array was set to 3 MHz. Each element was a circle with a diameter of 1.7 mm. The spacing between adjacent elements was 0.2 mm. The aperture diameter of the entire array was 30 mm. Data collection was performed at each array position for 10 average values ​​to ensure data stability and accuracy; Use a Gaussian window function centered on the array center frequency for filtering and Hilbert transform to process the time domain signal, with a 40dB half bandwidth of 90% of the array center frequency; The time domain signal of each transmit-receive element pair is captured and stored to form an FMC data set.

[0018] Preferably, in step S2, processing each FMC data set using a single-medium total focusing TFM algorithm comprises: Setting a size threshold of an imaging grid and creating an imaging grid whose size is larger than the set threshold; Calculate the intensity of each image point P1 ; Determine the location of the global maximum amplitude as the first point of the surface; Perform two passes, the first pass with the main direction being x and the secondary direction being y; the second pass with the main direction being y and the secondary direction being x. Finally, average the two passes to obtain two surface contour estimates. and .

[0019] Preferably, performing surface verification in step S2 to obtain the final three-dimensional surface profile of the target object includes: The final surface profile to be extracted Compared with the real surface generated by the Gaussian function equation, the z position error of each surface point is obtained to obtain the final extracted 3D surface profile.

[0020] Preferably, in step S3, calculating a new time of flight TOF, determining the ray path based on the Fermat shortest time principle, calculating the image intensity of the internal points, and obtaining a complete internal imaging of the target image includes: For each array position, a subset of extracted surface points is considered to calculate the TOF; The Fermat shortest time principle is used to determine the two surface intersection points that produce the global minimum total TOF; Calculate the distance from the extracted surface point to any image point TOF and store the TOF from each surface point into an array element TOF; Determine from arrive The true projection path represents the actual ultrasonic path, thereby obtaining a new TOF set.

[0021] Preferably, in step S3, calculating the image intensity of the internal point includes: This is accomplished by summing the intensity of all transmitting and receiving elements to calculate the intensity of each internal image point in the imaging grid. ; thus obtaining a 3D TFM internal image of each array position.

[0022] Preferably, in step S4, stitching all the internal TFM images together to create a three-dimensional TFM image of the entire interior of the sample includes: Input 3D TFM interior images from multiple array positions and create a 3D TFM image of the entire sample interior by stitching all interior TFM images together, resulting in a complete 3D TFM interior image.

[0023] Preferably, after obtaining the final three-dimensional surface contour and internal image reconstruction of the complete target object in step S4, the obtained results are analyzed, specifically including: The final extracted 3D surface profile imaging is evaluated, the z position error between the extracted surface and the real surface is compared, and the RMS error and other statistical indicators are calculated. The internal imaging results are evaluated, the position of visible defects is measured, their depth and lateral distance are compared, and the imaging accuracy is evaluated. The stitched 3D surface TFM image is displayed as an isosurface with color indicating depth, and the stitched 3D internal TFM image is displayed as an isosurface with color indicating distance.

[0024] Embodiment 2: This embodiment provides a control method for improving the detection accuracy of cable seal defective parts, and the method includes the following steps: like Figure 2 As shown, step S1: scanning the test sample to obtain a full matrix capture FMC data set: Specifically, the acquisition of the Full Matrix Capture (FMC) dataset is completed through a series of steps, as follows: (1) This embodiment uses a 2D sparse array containing 128 elements, which are arranged in a Poisson disk distribution. Each element is a circle with a diameter of 1.7 mm, the spacing between adjacent elements is about 0.2 mm, and the aperture diameter of the entire array is 30 mm. The operating frequency of the array is 3 MHz, and the -6 dB bandwidth at the center frequency is about 1.5 MHz.

[0025] (2) The test sample is a lead seal with a double-curved axisymmetric surface, and its surface morphology follows the Gaussian function: ; in is the peak height, is the standard deviation parameter, the steepest inclination angle is about 24 degrees, and the longitudinal propagation speed of sound in the sample v 2 is measured as 6360 m / s, so the wavelength at the array center frequency is is 2.1 mm.

[0026] (3) Perform a two-dimensional scan of the sample surface and collect data with an average value of 10 times at each array position to ensure data stability and accuracy. Use a Gaussian window function centered on the array center frequency for filtering and Hilbert transform to process the time domain signal, with a 40 dB half bandwidth of 90% of the array center frequency; the time domain signal of each transmitting-receiving element pair, i.e., the A-scan signal, is captured and stored to form an FMC data set; The specific formula for filtering the above process Gaussian window function is as follows: ; Where t is the time variable, is the standard deviation parameter, which determines the width of the window; for a 40 dB half-bandwidth of 90%, it can be calculated using the following relationship: ; where T is the period corresponding to the center frequency, and is the inverse function value of the error function, which is approximately 1.65; The specific formula of Hilbert transform in the above process is: ; Analyzing the signal It can be expressed as: ; Where j is an imaginary unit, the envelope of the analytical signal can be expressed as: ; The specific process of the above process is as follows: First, the data collected by the two-dimensional phased array is processed by the Gaussian window function on the original time domain signal. After weighting, the filtered signal is and ; The filtered signal Perform Hilbert transform to obtain the analytical signal With envelope signal ; The processed time domain signals, including the original signal, the filtered signal, and the envelope of the analyzed signal, are stored as part of the FMC data set; the FMC data set contains the time domain signals of all transmit-receive element pairs, namely the A-scan signals; for a 128-element 2D array, the FMC data set will include 128×128=16,384 A-scan signals, each of which represents an ultrasonic transmission and reception process from a specific transmitting element to a specific receiving element.

[0027] like Figure 3 As shown, in step S2, each obtained FMC data set is processed using a single-medium total focusing TFM algorithm to obtain a preliminary surface profile and a final surface profile of the target object. The specific process is as follows: After processing in S1, an FMC data set is obtained, and the FMC data set is used as an input signal in step S2; (1) Preliminary surface extraction: The FMC dataset is used as the input signal and the single-medium TFM algorithm is used. Since the sample is immersed in water for scanning in this scheme, the speed of sound in water needs to be considered in this algorithm. v1. Create an imaging grid large enough to cover the potential surface area. To create an imaging grid large enough and assuming that the surface is located at the location of the first reflection in the time domain data, the intensity of each image point P1 needs to be calculated. , which is calculated by taking into account the time delays of all pairs of transmitting and receiving elements: ; in, represents an optional weighting term, , no additional weight is applied; represents the complex Hilbert transform of the A-scan signal transmitted from the transmitting element T and received by the receiving element R; and are the position vectors of the transmitting and receiving elements respectively; represents the Euclidean norm is the speed of sound in water; After the above step (1), the intensity of each image point P1 is obtained. , this intensity value reflects the cumulative effect of all ultrasonic signals received at that point, which is based on the time delay of all transmit-receive paths.

[0028] (2) To determine the preliminary surface point, it is necessary to find the point with the maximum amplitude in the entire imaging grid; this is because surface reflections usually produce the strongest signal in the time domain data; the formula for finding the maximum amplitude point is as follows: ; in, is a point in the discrete surface imaging grid The image amplitude at is: ; Indicates the selection of the parameter value that makes the function take the maximum value, which refers to the coordinates here ; Through the above formula, we can get the point with the largest amplitude in the entire 3D TFM image. As a starting point for the preliminary surface extraction process; (3) After determining the global maximum amplitude point, As the starting point for surface extraction, starting from this point, surface points are gradually extracted along the main direction (x-axis) and the secondary direction (y-axis) until the entire surface area is covered. The specific steps are as follows: Initialization: The Z coordinate z k As a preliminary surface point ; Expand along the main direction: from Start by gradually expanding outward along the main direction (such as the x-axis) using spline fitting Estimate the coordinates of the neighboring points, for each point , check the k index that satisfies the following conditions, the formula is: ; Find the spline function The z coordinate that reaches the maximum value is used as the z coordinate of the next point. The function formula is: ; Handling missing points: If no points satisfying the condition are found in the range, slightly increase to try to cross small surface gaps; if 10 missing points are encountered in a row (corresponding to about 6 wavelengths), the surface edge is considered to have been reached; Expand along the secondary direction: For each newly found surface point, repeat the above process along the secondary direction (y-axis) to ensure full coverage of the surface; after the above steps, a preliminary set of surface points is obtained These points form a preliminary surface profile; this preliminary profile will be further refined in subsequent steps to obtain a more accurate final surface profile.

[0029] (4) The output obtained through steps (1)-(3) is the preliminary surface point set This is used as the input in this step; swap the primary and secondary directions, that is, first along the y direction and then along the x direction, repeat the above extraction process to obtain the second surface estimate ; Then the two surface estimates are averaged by the following formula: ; Thus outputting the preliminary surface profile ; (5) The above preliminary surface profile With the FMC dataset as input, in order to accurately calculate the propagation time of ultrasound in two media, it is necessary to consider the sound speed v1 in water and the sound speed v2 in the material; according to the Fermat principle, the ultrasound path is selected to minimize the total flight time. If the path involves two medium conversions, that is, from water to material and then from material to water, the following formula is used to determine the minimum total flight time TOF: ; For any image point P, assume its coordinates are , emitting element and accept elements The position coordinates are and ; and These are the interface points where the ultrasonic wave enters the material from the water and returns from the material to the water. and Found by solving the extreme value problem of Fermat's principle; (6) Based on the preliminary surface profile and the required imaging range, define a three-dimensional imaging grid that covers the entire area to be measured; add the signal values ​​of all transmit-receive paths to obtain the final image intensity at that point: ; in represents the complex Hilbert transform of the A-scan signal transmitted from the transmitting element T and received by the receiving element R; is the minimum total flight time from the transmitting element T to the image point P and then to the receiving element R; (7) Repeat the above steps: Repeat the surface extraction process in step (1) based on the more accurate TOF calculation until all image points have been processed to form a complete 3D TFM image and obtain the final surface profile. .

[0030] Step S3: Perform surface verification again to obtain the final three-dimensional surface contour of the target object. The specific steps are as follows: The final surface profile obtained in the previous step S2 Compared with the design surface equation, in this scheme the design surface equation is described by the following Gaussian function: ; Calculate the error at each surface point: To compare the final surface profile with the designed surface, calculate the actual height at each discrete surface point Design height The difference between: ; is the actual height value in the final surface profile, It is the height value calculated according to the design surface equation; The following statistical indicators are calculated: Minimum root mean square error ; Average Error: ; Through the above steps, the final surface profile can be fully evaluated The degree of conformity with the designed surface.

[0031] Step S4: Calculate the new flight time, use the Fermat shortest time principle to determine the ray path, calculate the image intensity of the internal points, and thus obtain the complete internal imaging of the target image: The input signal in S104 is a high-precision surface point set and FMC data set obtained through the preliminary and final surface extraction processes; in order to reduce calculation time and improve practicality, for each array position, only surface points located within a certain distance perpendicular to the array centerline, that is, within a cylinder relative to the array axis, are considered; these surface points will be used for subsequent TOF calculations.

[0032] According to Fermat's principle, the ultrasonic path is chosen to minimize the total flight time; for any internal image point P 2， Need to find the transmitting or receiving element The shortest path to P2; Represents transmission and reception respectively. represents the Euclidean norm, and the formula for TOF calculation is: ; In order to find the path that minimizes the total flight time, we need to find the surface interface point that minimizes the above TOF. and This represents the actual path of light rays on their way to or from this element: ; For each surface point to the internal image point P2 TOF and from each surface point to the array element The TOF is calculated and the result is stored. Based on the stored TOF information, the image intensity at each internal image point P2 can be calculated: ; By repeating the above calculations for all internal image points P2, the image intensity I(P2) can be synthesized throughout the entire internal imaging grid, thereby generating a complete internal 3D TFM image.

[0033] Step S5: Perform image stitching to stitch all internal TFM images together to create a three-dimensional TFM image of the entire sample interior. The specific method is as follows: 1) Input multiple partial 3D TFM images; 2) First, define a global coordinate system so that all partial 3D TFM images can be aligned and stitched in this coordinate system. Usually, this global coordinate system takes the first array position as the reference point; 3) To ensure that all partial images can be correctly stitched together, each partial image needs to be transformed into a global coordinate system based on the array position and scanning path; 4) In the overlapping area, select the pixel value with the largest amplitude as the value in the final image. This method is suitable for reducing the influence of noise and highlighting strong reflection features. The corresponding formula is: ; is the intensity value of point P in the final image, is the intensity value of point P in the i-th partial image; 5) For non-overlapping areas, directly copy the pixel values ​​in each partial image to the corresponding position in the final image; 6) Through the above steps, all partial 3D TFM images are stitched into a complete 3D image, covering the internal structure of the entire sample; the final output image not only contains the information of all scanning positions, but also ensures the continuity and consistency of the image.

[0034] S6, Detailed description of the imaging process: In MATLAB, N internal 3D TFM images are stitched into a complete 3D TFM image representing the internal structure of the sample. Similarly, the maximum amplitude method or other appropriate methods are used in the overlapping area to ensure image consistency and smooth transition. The corresponding smoothing technique formula is: ; in The input location is Strength value is the corresponding weight factor, which is defined based on the distance; finally, a continuous and smooth 3D image is output. Because the resolution is increased when reconstructing the image, and the reconstruction is performed for the entire volume, the corresponding imaging of the defective part of the cable seal is clearer.

Claims

1. A method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm, characterized in that: include: S1, the cable seal is scanned by a two-dimensional phased array probe to obtain a full matrix capture FMC data set; S2, using the single-medium full-focus TFM algorithm to process each obtained FMC data set to obtain the preliminary surface profile and the final surface profile of the target object; Perform surface verification to obtain the final three-dimensional surface profile of the target object; S3, calculate the new flight time TOF, determine the ray path based on the Fermat shortest time principle, calculate the image intensity of the internal points, and obtain the complete internal imaging of the target image; S4, image stitching is performed to stitch all internal TFM images together to create a three-dimensional TFM image of the entire interior of the sample, and ultimately obtain the final three-dimensional surface contour and internal image reconstruction of the complete target object.

2. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 1 is characterized in that: In step S1, the cable seal is scanned to obtain a full matrix capture FMC data set including: Construct a 2D sparse array of N elements, arranged in a Poisson disk distribution to avoid grating effects and maintain high resolution; set the array's operating frequency, the diameter of each circular element, the spacing between adjacent elements, and the aperture diameter of the entire array; Perform M average data collection at each array position to ensure data stability and accuracy; Process the time domain signal using a Gaussian window function centered on the array center frequency for filtering and Hilbert transform; The time domain signal of each transmit-receive element pair is captured and stored to form an FMC data set.

3. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 1 is characterized in that: In step S2, processing each FMC data set using the single-medium total focusing TFM algorithm includes: Setting a size threshold of an imaging grid and creating an imaging grid whose size is larger than the set threshold; Calculate the intensity of each image point P1 ; Determine the location of the global maximum amplitude as the first point of the surface; Perform two passes, the first pass with the main direction being x and the secondary direction being y; the second pass with the main direction being y and the secondary direction being x. Finally, average the two passes to obtain two surface contour estimates. and .

4. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 3 is characterized in that: Surface verification is performed in step S2 to obtain the final three-dimensional surface profile of the target object, including: The final surface profile to be extracted Compared with the real surface generated by the Gaussian function equation, the z position error of each surface point is obtained to obtain the final extracted 3D surface profile.

5. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 4 is characterized in that: In step S3, a new time of flight TOF is calculated, the ray path is determined based on the Fermat shortest time principle, the image intensity of the internal points is calculated, and the complete internal imaging of the target image is obtained, including: For each array position, a subset of extracted surface points is considered to calculate the TOF; The Fermat shortest time principle is used to determine the two surface intersection points that produce the global minimum total TOF; Calculate the distance from the extracted surface point to any image point TOF and stores the TOF from each surface point into an array element TOF; Determine from arrive The true projection path represents the actual ultrasonic path, thereby obtaining a new TOF set.

6. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 5 is characterized in that: In step S3, calculating the image intensity of the internal points includes: This is accomplished by summing the intensity of all transmitting and receiving elements to calculate the intensity of each internal image point in the imaging grid. ; thus obtaining a 3D TFM internal image of each array position.

7. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 1 is characterized in that: In step S4, all internal TFM images are stitched together to create a three-dimensional TFM image of the entire interior of the sample including: Input 3D TFM interior images from multiple array positions and create a 3D TFM image of the entire sample interior by stitching all interior TFM images together, resulting in a complete 3D TFM interior image.

8. The method for improving the detection accuracy of cable seal defective parts based on full matrix capture and full focusing algorithm according to claim 1 is characterized in that: After obtaining the final three-dimensional surface contour and internal image reconstruction of the complete target object in step S4, the obtained results are analyzed, specifically including: The final extracted 3D surface profile imaging is evaluated, the z position error between the extracted surface and the real surface is compared, and the RMS error and other statistical indicators are calculated. The internal imaging results are evaluated, the position of visible defects is measured, their depth and lateral distance are compared, and the imaging accuracy is evaluated. The stitched 3D surface TFM image is displayed as an isosurface with color indicating depth, and the stitched 3D internal TFM image is displayed as an isosurface with color indicating distance.

Images