Weld joint parameter anti-interference measurement method

Abstract

The invention provides a welding seam parameter anti-interference measurement method, which comprises the following steps of: acquiring spectrum characteristics of a high-reflectivity surface through a laser scattering technology and Fourier transform processing, matching a correction coefficient, realizing weighted adjustment and geometric parameter extraction of a scattering signal, accurately quantifying a groove size in combination with an edge enhancement algorithm and a three-dimensional contour modeling technology, and accurately measuring the welding seam parameters. Interference such as surface oil stain, corrosion and stripe fracture is effectively overcome by using image filtering, morphological reconstruction and an abnormal value elimination algorithm, and finally, high-precision and anti-interference measurement of the weld groove angle, the truncated edge thickness and the assembly alignment tolerance under high-reflection, corrosion and complex working conditions is realized through multi-source data fusion and self-adaptive calculation range adjustment. And the reliability and the automation level of welding quality detection are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pipe welding, in particular to a welding seam parameter anti-interference measurement method. BACKGROUND

[0002] In actual production, when collecting the three-dimensional appearance data of the workpiece groove by a three-dimensional scanner, the process of obtaining the groove face angle and the land thickness faces a complex technical contradiction.

[0003] Specifically, when the workpiece material is a high-reflectivity metal and the surface has fine oil stains, the light reflection signal of the three-dimensional scanner will become unstable due to the material characteristics and oil stain interference, resulting in a ±2° deviation in the measurement of the groove face angle, and the land thickness data will be magnified to more than 0.5 millimeters due to the blurred boundary.

[0004] Then, in the pipe assembly and pairing process, when measuring the misalignment data and the groove size after assembly and pairing by using the structured light scanning technology, if there is local rust or slight deformation on the surface of the pipe, the structured light stripe projection will be broken due to the change of the surface texture, resulting in a deviation of 1 millimeter between the measured value and the actual value. The technical contradiction lies in the mutual restriction between the high-precision scanning requirement and the surface state under complex working conditions, which leads to the difficulty in maintaining the continuity and reliability of the whole process data chain from groove preparation, assembly and pairing to post-welding detection in special scenarios. SUMMARY

[0005] The first aspect of the present disclosure provides a welding seam parameter anti-interference measurement method, comprising: obtaining the scattering light distribution data of the high-reflectivity surface by using the laser scattering technology, performing Fourier transform processing on the scattering light distribution data to obtain the frequency spectrum characteristics; matching the correction coefficient in the pre-established model according to the frequency spectrum characteristics, weighting and adjusting the scattering light distribution data by using the correction coefficient to generate a correction signal; extracting the initial measurement value of the geometric parameter from the correction signal, adjusting the initial measurement value by using smoothing processing to obtain geometric data; separating the boundary feature signal from the correction signal, processing the boundary feature signal by using an edge enhancement algorithm to generate boundary enhancement data; constructing a three-dimensional contour model by using the boundary enhancement data and the geometric data, performing quantitative calculation on the three-dimensional contour model by using segmentation technology to obtain quantitative data; obtaining the original image of the projection data, denoising the original image by using filtering processing to generate a filtered image; extracting the continuity feature from the filtered image, processing the continuity feature to generate repair data; According to the repair data, a deviation amount is calculated by combining a triangulation principle, the deviation amount is processed by using an outlier rejection algorithm, and deviation data is obtained; Depth map data is acquired, the depth map data is processed by classification segmentation, a calculation range is adjusted by combining the deviation data and the quantization data, and quantization data is generated.

[0006] In combination with the first aspect, the scattering light distribution data of the high-reflectivity surface is acquired by the laser scattering technology, the scattering light distribution data is processed by Fourier transform, and frequency spectrum features are obtained, including: The initial light distribution data of the high-reflectivity surface is collected by the laser scattering technology, the distribution patterns of the scattering light at different angles are recorded, and an initial light distribution data set is obtained; The scattering light intensity information is extracted from the initial light distribution data set, the light intensity values at each angle are normalized, and scattering light intensity sequence data are obtained; The fast Fourier transform algorithm is used on the scattering light intensity sequence data, and a frequency spectrum data set is generated; The frequency spectrum peak value and the frequency distribution features are extracted from the frequency spectrum data set, the significant feature points are judged according to a preset threshold, and a frequency spectrum feature set is obtained; The frequency spectrum feature set is processed by the principal component analysis algorithm, and a feature vector set after dimension reduction is obtained; The support vector machine classifier is used to classify and train the feature vector set after dimension reduction, a classification model is generated, and surface characteristic judgment data are obtained.

[0007] In combination with the first aspect, the correction coefficient is matched in a pre-established model according to the frequency spectrum features, the scattering light distribution data is weighted and adjusted by using the correction coefficient, and a correction signal is generated, including: The statistical feature values of the distribution data are obtained by using the frequency spectrum features, the mean value and the variance calculation method are used, and a statistical feature set is obtained; The amplitude value information corresponding to the significant frequency is extracted from the statistical feature set, the amplitude value significance is screened according to a preset threshold, and an amplitude feature set is obtained; The deviation value is calculated according to the amplitude feature set and the standard amplitude data in the pre-established model, the mapping relationship between the deviation value and the correction coefficient is determined, and the correction coefficient combination is determined; The scattering light distribution data is weighted and adjusted by using the correction coefficient combination, and a correction signal data set is obtained; The peak value offset of the frequency distribution is extracted from the correction signal data set, the support vector machine classifier is used to judge the offset distribution mode, and an offset classification result is obtained; The offset classification result is processed by the principal component analysis algorithm, and a feature vector set is obtained.

[0008] In combination with the first aspect, the initial measurement value of the geometric parameter is extracted from the modified signal, the initial measurement value is adjusted by a smoothing process to obtain geometric data, including: The spectral distribution characteristics of the geometric parameter are obtained from the modified signal, and the spectral distribution is decomposed by a fast Fourier transform to obtain a spectral feature set; The significant frequency components related to the geometric parameter are extracted from the spectral feature set, and the frequency significance is filtered according to a preset threshold to obtain a frequency feature set; The frequency deviation value is calculated according to the frequency feature set and the standard frequency data in the preset constraint, and the adjustment coefficient is determined by using the mapping relationship between the deviation value and the geometric parameter; The initial measurement value is adjusted by the adjustment coefficient to obtain the geometric data; If the deviation of the geometric data from the preset constraint exceeds the preset threshold, the geometric data is smoothed by a Gaussian filtering method to obtain the geometric data; The statistical features of the distribution are extracted from the geometric data, and the statistical feature set is obtained by using the mean and standard deviation calculation method.

[0009] In combination with the first aspect, the boundary feature signal is separated from the modified signal, and the boundary feature signal is processed by using an edge enhancement algorithm to generate boundary enhancement data, including: The boundary feature signal is separated from the modified signal by using a signal decomposition method, and the frequency domain analysis of the boundary feature signal is performed by using a wavelet transform technique to obtain a frequency domain feature set; The significant boundary frequency components are extracted from the frequency domain feature set, and the frequency significance is filtered according to a preset threshold to obtain a boundary frequency set; The boundary enhancement set is obtained by performing convolution operation on the boundary frequency set by using a two-dimensional Gaussian kernel; According to the boundary enhancement set and the boundary feature signal, the principal component analysis is used to perform dimension reduction processing on the enhanced data to obtain feature dimension reduction data; If the deviation of the feature dimension reduction data from the preset boundary constraint exceeds the preset threshold, the feature dimension reduction data is smoothed by a Gaussian filtering method to obtain a smoothed feature set; The boundary distribution statistical features are extracted from the smoothed feature set, and the statistical feature set is obtained by using the mean and variance calculation method.

[0010] In combination with the first aspect, the three-dimensional contour model is constructed by using the boundary enhancement data and the geometric data, and the three-dimensional contour model is quantitatively calculated by using a segmentation technique to obtain quantitative data, including: The boundary enhanced data and the geometric data are used to perform spatial discretization processing on the target region by using a triangular mesh division technique to generate a discrete mesh set; A curvature value of each mesh vertex is extracted from the discrete mesh set to form a surface curvature feature; The curvature feature is subjected to a smoothing operation by using a Gaussian filter to obtain a curvature feature set; The curvature feature set is subjected to principal component analysis to compress the feature dimension to three dimensions to generate a reduced dimension feature set; If the standard deviation of the reduced dimension feature set exceeds a preset threshold, the reduced dimension feature set is subjected to smoothing adjustment by using mean filtering to obtain a smoothed feature set; According to the smoothed feature set, a K-means clustering algorithm is used to group the feature points to generate a grouped feature set.

[0011] In combination with the first aspect, the continuity feature is extracted from the filtered image, and a repair algorithm is used to process the continuity feature to generate repair data, including: The continuity feature is extracted from the filtered image, and continuity quantization data is obtained by calculating the continuity length and the spacing ratio; If the continuity quantization data is lower than a preset threshold, a morphological reconstruction algorithm is used to perform dilation and corrosion operations on the broken region to generate preliminary repair data; The preliminary repair data is used to extract a boundary point set of the feature by using a Canny edge detection algorithm to obtain boundary feature data; The boundary feature data is subjected to Hough line transformation to detect the linear distribution characteristics to generate a linear distribution feature set; The distribution density of the broken region is extracted from the linear distribution feature set, and the distribution density is subjected to smoothing processing by using median filtering to obtain smoothed density data; According to the smoothed density data, a K-nearest neighbor algorithm is used to classify the feature points of the broken region to generate a classified feature point set; The classified feature point set is matched with a pre-established template database to obtain repair data.

[0012] In combination with the first aspect, the deviation amount is calculated according to the repair data in combination with the triangulation principle, and an outlier rejection algorithm is used to process the deviation amount to obtain deviation data, including: Position and spatial information is extracted from the repair data, and an initial deviation amount value is calculated by using triangulation to generate initial deviation amount value data; The initial deviation amount value data is subjected to an isolation forest algorithm to detect outliers, and an outlier point is determined by calculating an outlier score, and deviation data is generated after the outlier point is rejected; Obtain the spatial distribution characteristics of the deviation value from the deviation data, smooth the distribution data using a mean filter to obtain smoothed deviation distribution data; According to the smoothed deviation distribution data, use the DBSCAN algorithm to cluster the feature points of the deviation region to generate a cluster feature point set; Match the cluster feature point set with a pre-established distribution template database, calculate the distance deviation of each feature point from the template using the Euclidean distance, and determine the optimized deviation distribution data; Obtain the local density data from the optimized deviation distribution data, smooth the density data using a median filter to obtain the final deviation value data.

[0013] In combination with the first aspect, the method further comprises, after generating the quantization data: Compare the quantization data with a preset groove size standard range, and if it exceeds the standard range, generate an alarm signal and output an adjustment suggestion.

[0014] The second aspect of the present disclosure provides an electronic device comprising: One or more processors; A storage unit for storing one or more programs, which when executed by the one or more processors, can cause the one or more processors to implement the method.

[0015] Beneficial effects: The welding seam parameter anti-interference measurement method provided by the present application realizes the weighted adjustment of the scattering signal and the extraction of geometric parameters by obtaining the frequency spectrum characteristics of the high-reflectivity surface and matching the correction coefficient through laser scattering technology and Fourier transform processing, accurately quantizes the groove size by combining the edge enhancement algorithm and the three-dimensional contour modeling technology, and effectively overcomes the interference of surface oil stains, rust and stripe fracture by using image filtering, morphological reconstruction and outlier rejection algorithm. Finally, through multi-source data fusion and adaptive calculation range adjustment, high-precision, anti-interference measurement of welding seam groove angle, blunt edge thickness and group error is realized under high-reflectivity, rust and complex working conditions, which significantly improves the reliability and automation level of welding quality detection. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a welding seam parameter anti-interference measurement method according to an embodiment of the present disclosure; Figure 2 An electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0017] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, like reference numerals refer to like elements, unless the context clearly dictates otherwise. The following description of exemplary embodiments is not representative of all possible embodiments consistent with the present disclosure.

[0018] The terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0019] As shown in FIG. 1, a flowchart of a weld parameter anti-interference measurement method according to an embodiment of the present disclosure is shown, which includes the following steps: Figure 1 S101: Obtain the scattering light distribution data of the high-reflectivity surface by using the laser scattering technology, and perform Fourier transform processing on the scattering light distribution data to obtain the frequency spectrum features; S101: Obtain the scattering light distribution data of the high-reflectivity surface by using the laser scattering technology, and perform Fourier transform processing on the scattering light distribution data to obtain the frequency spectrum features; S101: Obtain the scattering light distribution data of the high-reflectivity surface by using the laser scattering technology, and perform Fourier transform processing on the scattering light distribution data to obtain the frequency spectrum features; Collect the initial light distribution data of the high-reflectivity surface by using the laser scattering technology, record the distribution patterns of the scattering light at different angles, and obtain an initial light distribution data set; Extract the scattering light intensity information from the initial light distribution data set, normalize the light intensity value at each angle, and obtain the scattering light intensity sequence data; Perform fast Fourier transform algorithm on the scattering light intensity sequence data to generate a frequency spectrum data set; Extract the frequency spectrum peak value and the frequency distribution features from the frequency spectrum data set, judge the significant feature points according to the preset threshold, and obtain a frequency spectrum feature set; Perform dimensionality reduction processing on the frequency spectrum feature set by using the principal component analysis algorithm to obtain a feature vector set after dimensionality reduction; Perform classification training on the feature vector set after dimensionality reduction by using the support vector machine classifier to generate a classification model, and obtain the surface characteristic determination data.

[0020] Specifically, in step S101, the scattering light distribution data of the high-reflectivity surface is obtained by using the laser scattering technology, and the frequency spectrum features representing the surface characteristics are extracted by performing Fourier transform processing on the data.

[0021] In a specific implementation, first, a laser scattering technique is used to scan a highly reflective metal surface. After the laser beam is irradiated onto the workpiece surface, it will produce a unique scattering distribution due to the differences in surface micro-topography, material, and attachments (such as oil stains). By collecting the scattering light intensity at different angles through high-precision sensors, an initial light distribution data set containing spatial and intensity information is formed.

[0022] Subsequently, the core light intensity information is extracted from the data set. In order to eliminate the dimensional influence brought by environmental lighting and equipment differences, the original light intensity values collected at each angle are normalized to convert them to a unified scale, obtaining a stable and comparable sequence of scattering light intensity data.

[0023] Next, in order to reveal the periodicity or characteristic pattern contained in the scattering signal, the Fast Fourier Transform (FFT) algorithm is used to convert the light intensity sequence in the time or space domain to the frequency domain. After FFT processing, a frequency spectrum data set is generated, which contains amplitude and phase information representing different surface characteristic frequency components.

[0024] Then, from the complex frequency spectrum data set, the key spectral peaks and specific frequency distribution characteristics are automatically identified and extracted by an algorithm.

[0025] According to the threshold value calibrated in advance through a large number of experiments, the system can intelligently determine which are the significant feature points representing the surface material, oil stains, and other states, thereby filtering out the effective components that constitute the frequency spectrum feature set.

[0026] Since the initial frequency spectrum feature set has a high dimension and may have redundancies, in order to further condense information and highlight the main features, a Principal Component Analysis (PCA) algorithm is used to reduce the dimension of the frequency spectrum feature set. This process converts the original features to a new set of dimensions with fewer numbers, mutual independence, and maximum information content, obtaining a reduced feature vector set.

[0027] Finally, a machine learning method is used to enable the system to intelligently determine the surface characteristics. A Support Vector Machine (SVM) classifier is used to supervise the learning and training of the reduced feature vector set, generating a classification model that can accurately classify different surface states (such as clean metal, with oil stains, or slight rust).

[0028] By applying this model, new scattering light data can be judged, and the final determination data representing the surface characteristics can be output, providing a key basis for subsequent signal correction.

[0029] S102: According to the matching of the frequency spectrum features in the pre-established model, a correction coefficient is matched, and the scattering light distribution data is weighted and adjusted using the correction coefficient to generate a corrected signal; The correction coefficient is matched in a pre-established model according to the spectral characteristics, the scattering light distribution data is weighted and adjusted by using the correction coefficient, and a correction signal is generated, comprising: The statistical characteristic values of the distribution data are obtained through the spectral characteristics, and the statistical characteristic set is obtained by using the mean value and variance calculation method. The amplitude information corresponding to the significant frequency is extracted from the statistical characteristic set, and the amplitude characteristic set is obtained by screening the amplitude significance according to the preset threshold. The deviation value is calculated according to the amplitude characteristic set and the standard amplitude data in the pre-established model, and the correction coefficient combination is determined through the mapping relationship between the deviation value and the correction coefficient. The scattering light distribution data is weighted and adjusted by using the correction coefficient combination, and a correction signal data set is obtained. The peak shift of the frequency distribution is extracted from the correction signal data set, the shift distribution pattern is judged by the support vector machine classifier, and the shift classification result is obtained. The principal component analysis algorithm is used to reduce the dimension of the shift classification result, and the characteristic vector set is obtained.

[0030] Specifically, in step S102, the system matches the most suitable correction coefficient in the pre-established expert model according to the spectral characteristics obtained in the foregoing steps, and uses these coefficients to weight and adjust the original scattering light distribution data, and finally generates a correction signal with stronger anti-interference ability. This process is the key link to realize accurate measurement, aiming to eliminate or weaken the signal distortion caused by high reflectivity and pollution of the surface.

[0031] The specific implementation path is as follows: first, the system will conduct in-depth statistical analysis on the obtained spectral characteristics. By calculating the mean value, variance and other statistical characteristic values, the complex spectral information is converted into a set of statistical characteristic sets that can quantify the distribution characteristics, laying a data foundation for subsequent accurate matching.

[0032] Next, the algorithm focuses on the frequency components that are most sensitive to the surface state. It extracts the amplitude information corresponding to the significant frequency from the statistical characteristic set, and screens out the amplitude points with significant according to the preset threshold, so as to condense the core amplitude characteristic set. This step ensures that the subsequent correction is aimed at the most critical signal-to-noise points.

[0033] Subsequently, the system starts the core matching calculation. It compares the amplitude characteristic set obtained by the current measurement with the standard amplitude data in the standard model library established by a large amount of experimental data, calculates the deviation value between them. Each deviation value is converted into a specific correction coefficient through a preset mapping relationship function, and finally forms a correction coefficient combination for the current surface state.

[0034] After obtaining the revised coefficient combination, the system applies it to the original scattered light distribution data. This process is an adaptive weighted adjustment, where each data point is given a different weight according to its corresponding frequency characteristics, generating an accurately calibrated revised signal dataset. This effectively suppresses the interference signal and enhances the representation of the true surface topography information.

[0035] To further ensure the quality of the revision, the system performs a closed-loop test on the revised result. It extracts the peak shift and other features of the frequency distribution from the revised signal dataset and uses a pre-trained support vector machine (SVM) classifier to intelligently judge the distribution pattern of these shifts, obtaining a classification result about the shift pattern to evaluate the revision effect.

[0036] Finally, to make the data more suitable for subsequent processing, the system uses principal component analysis (PCA) algorithm to reduce the dimension of the above shift classification result, eliminate redundant information, condense core features, and output a refined feature vector set. This set not only represents the core information of the revised signal, but also serves as a high-quality input for subsequent three-dimensional reconstruction and geometric parameter calculation.

[0037] S103: Extracting initial measurement values of geometric parameters from the revised signal, adjusting the initial measurement values through smoothing processing to obtain geometric data; The method for extracting initial measurement values of geometric parameters from the revised signal and adjusting the initial measurement values through smoothing processing to obtain geometric data includes: Obtaining the frequency spectrum distribution characteristics of the geometric parameters through the revised signal, decomposing the frequency spectrum distribution using fast Fourier transform to obtain a frequency spectrum feature set; Extracting significant frequency components related to the geometric parameters from the frequency spectrum feature set, filtering frequency significance according to a pre-set threshold to obtain a frequency feature set; Calculating frequency deviation values according to the frequency feature set and standard frequency data in the pre-set constraints, determining adjustment coefficients using the mapping relationship between the deviation values and the geometric parameters; Adjusting the initial measurement values through the adjustment coefficients to obtain geometric data; If the deviation between the geometric data and the pre-set constraints exceeds a pre-set threshold, smoothing the geometric data using a Gaussian filtering method to obtain geometric data; Extracting statistical features of the distribution from the geometric data, obtaining a statistical feature set using mean and standard deviation calculation methods.

[0038] In step S103, the system extracts initial measurement values reflecting the geometry of the bevel (such as angle, land thickness) from the anti-interference corrected signal, and optimizes these initial values through smoothing techniques, finally outputting high-precision, high-stability geometric data. The core purpose of this step is to convert the corrected signal into reliable three-dimensional topographic information.

[0039] The specific implementation path is as follows: First, the system performs frequency domain analysis on the corrected signal to deeply understand the geometric information it contains. Through Fast Fourier Transform (FFT), the signal's frequency spectrum distribution is decomposed to obtain a spectrum feature set containing various frequency components. Changes in geometric topography, such as the inclination of the bevel surface or the edge of the land, will exhibit obvious characteristics at specific frequencies.

[0040] Subsequently, the algorithm focuses on identifying key frequency components that are strongly related to geometric parameters. It extracts these significant frequency components from the spectrum feature set and filters them according to a pre-set significance threshold, thereby obtaining a frequency feature set that can most effectively represent the current geometric shape. This step ensures that subsequent calculations are based on the most relevant information.

[0041] Next, the system initiates a calibration process. It compares the current extracted frequency feature set with the standard frequency data (derived from an ideal model or calibration sample) in the pre-set constraints, and calculates the frequency deviation values. Each frequency deviation value is converted into an adjustment coefficient for a specific geometric parameter through a pre-established mapping relationship model.

[0042] After obtaining the adjustment coefficients, the system applies them to the initial measurement values directly calculated from the signal. This is a weighted adjustment process aimed at systematically correcting deviations caused by residual noise or model mismatches, thereby generating a set of preliminary optimized geometric data.

[0043] To ensure the reliability and smoothness of the data, the system performs a verification operation. It checks whether the set of geometric data meets the pre-set constraints (such as angle range, thickness reasonableness). If the deviation of the data from the constraints exceeds the pre-set tolerance threshold, the Gaussian filtering method is used to smooth the set of geometric data. Gaussian filtering can effectively suppress high-frequency noise and abnormal fluctuations in the data while well preserving the true geometric trend, thereby obtaining a more smooth and reasonable set of geometric data.

[0044] Finally, to quantitatively evaluate the quality of the final set of geometric data, the system extracts statistical features of its distribution, such as calculating the mean and standard deviation of all angle or thickness measurements. The obtained set of statistical features provides important reference information about the current data confidence and dispersion degree for subsequent steps (such as three-dimensional modeling).

[0045] S104: separate a boundary feature signal from the correction signal, process the boundary feature signal by using an edge enhancement algorithm, and generate boundary enhancement data; The separating a boundary feature signal from the correction signal, processing the boundary feature signal by using an edge enhancement algorithm, and generating boundary enhancement data include: The boundary feature signal is separated from the correction signal by using a signal decomposition method, and a frequency domain analysis is performed on the boundary feature signal by using a wavelet transform technique to obtain a frequency domain feature set; Significant boundary frequency components are extracted from the frequency domain feature set, and a boundary frequency set is obtained according to a preset threshold value for filtering frequency significance; A two-dimensional Gaussian kernel is used for convolution operation on the boundary frequency set to obtain a boundary enhancement set; According to the boundary enhancement set and the boundary feature signal, principal component analysis is used to perform dimension reduction processing on the enhancement data to obtain feature dimension reduction data; If the feature dimension reduction data deviates from a preset boundary constraint by more than a preset threshold value, a Gaussian filtering method is used to perform smoothing operation on the feature dimension reduction data to obtain a smoothed feature set; Boundary distribution statistical features are extracted from the smoothed feature set, and a statistical feature set is obtained by using a mean and variance calculation method.

[0046] In step S104, the system aims to solve the common boundary blur problem in blunt edge thickness measurement. The core task is to accurately separate the feature signal representing the physical boundary from the correction signal containing rich information, and to process it by using an advanced edge enhancement algorithm, and finally to generate a clear boundary and feature sharpening data, which lays a foundation for subsequent accurate quantitative calculation.

[0047] The specific implementation is a multi-stage fine processing process. First, the system uses a signal decomposition method to separate the feature signal component mainly bearing boundary information from the complex correction signal. Then, a wavelet transform technique is used to perform multi-resolution frequency domain analysis on the boundary feature signal. The wavelet transform can provide good localization characteristics in time domain and frequency domain at the same time, so as to accurately capture the features of boundary mutation in different scales, forming a frequency domain feature set containing multi-scale information.

[0048] Subsequently, the algorithm intelligently extracts the significant frequency components most sensitive to the boundary from the frequency domain feature set. By filtering through a preset significance threshold, the system focuses on the most critical information to obtain a pure boundary frequency set that enhances the boundary features.

[0049] To further sharpen the edges and suppress noise, the system performs a convolution operation on the boundary frequency set. This operation uses a two-dimensional Gaussian kernel as the convolution kernel, which has the effect of enhancing the continuity of the edges while smoothing out high-frequency noise and unnecessary details, thereby generating a boundary-enhanced set with significantly improved signal-to-noise ratio.

[0050] After that, the system combines the edge-enhanced data with the original boundary feature signal and uses principal component analysis (PCA) to reduce the dimensionality of the fused high-dimensional data. PCA effectively eliminates information redundancy and noise, condenses low-dimensional features that best represent the essential structure of the boundary, and obtains feature-reduced data.

[0051] To ensure that the generated boundary data meets the physical rationality and consistency, the system compares it with the pre-set boundary constraints (such as the expected boundary position, direction, etc.). If the deviation exceeds the allowed threshold, the Gaussian filtering method is used again to smooth the feature-reduced data to eliminate possible outliers and non-smooth jumps, thereby obtaining a more robust and smooth feature set.

[0052] Finally, the system extracts statistical features of the boundary distribution from the processed and smoothed feature set, such as quantifying the position and clarity of the boundary by calculating the mean and variance. The resulting statistical feature set not only quantitatively evaluates the processing effect of this step, but also provides high-quality boundary information input for the accurate construction of the three-dimensional contour model in the subsequent step.

[0053] S105: Construct a three-dimensional contour model from the boundary-enhanced data and the geometric data, use segmentation technology to quantitatively calculate the three-dimensional contour model, and obtain quantitative data; The construction of a three-dimensional contour model from the boundary-enhanced data and the geometric data, the use of segmentation technology to quantitatively calculate the three-dimensional contour model, and the obtaining of quantitative data include: Using the boundary-enhanced data and the geometric data, use the triangular mesh division technology to perform spatial discretization processing on the target region, and generate a discrete mesh set; Extract the curvature value of each mesh vertex from the discrete mesh set to form a surface curvature feature; Use Gaussian filtering to smooth the curvature feature to obtain a curvature feature set; For the curvature feature set, use principal component analysis to compress the feature dimension to three dimensions to generate a reduced dimension feature set; If the standard deviation of the reduced dimension feature set exceeds the pre-set threshold, use mean filtering to smooth the reduced dimension feature set to obtain a smoothed feature set; According to the smoothed feature set, use the K-means clustering algorithm to group the feature points to generate a grouped feature set.

[0054] In step S105, the system fuses the boundary enhancement data obtained in the previous steps with the geometric data, aiming to construct a digital model that accurately reflects the three-dimensional topography of the pipe bevel, and uses segmentation techniques to quantitatively analyze the model, finally outputting key size quantization data. This step is the core link of the conversion from the original signal to the final measurement index.

[0055] The specific implementation process is as follows: First, the system combines high-definition boundary data with data representing macro-geometric morphology, and uses triangular subdivision grid division technology to perform spatial discretization processing on the target bevel area. This process converts the continuous surface into a discrete geometric set composed of a large number of triangular grids, establishing a digital framework for the three-dimensional contour model and providing a basic topological structure for subsequent detailed analysis.

[0056] Subsequently, the algorithm extracts geometric properties representing the degree of surface curvature from the discrete grid set. By calculating the curvature value of each grid vertex, the system forms a set of surface curvature feature data that can finely describe the convexity, concavity, and turning features of the bevel surface. These data are key to identifying bevel surface angles and fillet regions.

[0057] To suppress noise fluctuations that may be introduced in curvature calculation and improve data stability, the system uses Gaussian filtering to smooth the original curvature features. After processing, a set of smoothed curvature features that better reflect the true topography trend is obtained.

[0058] In the face of possible multi-dimensional and redundant curvature data, the system uses principal component analysis (PCA) technology to compress the features, projecting the core features of the data into a three-dimensional space to generate a reduced dimension feature set that retains the main information and is easy to process.

[0059] The system continuously monitors data quality. If the standard deviation of the reduced dimension feature set exceeds the preset threshold, indicating that the data has abnormal fluctuations or outliers. At this time, the system adjusts the data through mean filtering to further eliminate uncertainty and obtain a more robust smoothed feature set.

[0060] Finally, based on this processed high-quality feature data, the system uses the K-means clustering algorithm to intelligently group all feature points. This algorithm can automatically aggregate points with similar geometric features (such as belonging to the same bevel surface or fillet) together, generating a clear grouped feature set. This process essentially segments the three-dimensional model into different feature regions, providing a direct basis for calculating bevel angle, fillet thickness, and other quantitative parameters.

[0061] S106: Obtain an original image of the projection data, denoise the original image through filtering processing, and generate a filtered image; The continuity feature is extracted from the filtered image, and a repair algorithm is used to process the continuity feature to generate repair data, including: The continuity feature is extracted from the filtered image, and the continuity quantization data is obtained by calculating the continuity length and spacing ratio; If the continuity quantization data is lower than the preset threshold, a morphological reconstruction algorithm is used to perform dilation and corrosion operations on the broken area to generate preliminary repair data; Through the preliminary repair data, a Canny edge detection algorithm is used to extract a boundary point set of the feature to obtain boundary feature data; For the boundary feature data, a Hough line transformation is used to detect the linear distribution characteristics to generate a linear distribution feature set; The distribution density of the broken area is extracted from the linear distribution feature set, and the distribution density is smoothed by median filtering to obtain smoothed density data; According to the smoothed density data, a K-nearest neighbor algorithm is used to classify the feature points of the broken area to generate a classified feature point set; Through the classified feature point set and the pre-established template database, repair data is obtained.

[0062] In step S106, the system aims at the problem of stripe projection breakage caused by local rust or deformation of the pipeline surface during the structured light scanning process, and finally generates complete stripe data by preprocessing and intelligent repair of the original projection image, providing reliable input for subsequent high-precision three-dimensional measurement.

[0063] The specific implementation process is a layer-by-layer progressive precision processing procedure. First, the system extracts features representing the continuity of the stripes from the filtered and denoised image. By calculating the ratio of the continuity length of the stripes to the spacing of the broken areas, the system obtains a set of quantifiable continuity index data, which provides a basis for objectively judging the severity of stripe breakage.

[0064] When the system detects that the continuity quantization data is lower than the preset safety threshold, it indicates that the stripes have obvious breakage and the repair program needs to be started. At this time, the system uses a morphological reconstruction algorithm to process the identified broken areas. Through a series of precise dilation and corrosion operations, the algorithm can intelligently fill the broken gap and eliminate noise interference according to the surrounding complete stripe structure features, thereby generating a preliminary repaired image data.

[0065] In order to accurately locate the edge profile of the repaired stripe, the system performs Canny edge detection on the preliminary repair data. The Canny algorithm, with its excellent signal-to-noise ratio and precise single-edge response characteristics, can extract a clear and coherent set of stripe boundary points, providing accurate boundary feature data for subsequent geometric analysis.

[0066] Due to the strong linear distribution characteristics of structured light stripes, the system then uses Hough line transformation to process the boundary feature data. Hough transformation can effectively detect straight lines in the image and associate edge segments that are broken but belong to the same straight line, thereby generating a feature set that describes the linear distribution characteristics of all stripes, such as direction and position.

[0067] Next, the system analyzes the distribution of the broken area from the linear distribution feature set and extracts its distribution density information. In order to eliminate possible impulse noise and abnormal fluctuations in the density data, the system uses median filtering to smooth it, obtaining a smoothed density data that more truly reflects the distribution of the broken area.

[0068] S107: Extracting continuity features from the filtered image, processing the continuity features, and generating repair data; In step S107, the system focuses on solving the problem of structured light stripe breakage due to surface rust or deformation, and generates complete stripe data that can be used for accurate measurement by extracting and intelligently repairing the stripe continuity features in the filtered image.

[0069] First, the continuity features of the stripes are extracted from the filtered image. The system generates quantitative continuity index data by calculating the connection degree between adjacent stripe segments, the width of the broken gap, and the continuity of the stripe direction. These data objectively reflect the integrity and quality of the stripe image.

[0070] When the system detects that the continuity index is below the preset threshold, it automatically triggers the repair process. The morphological reconstruction algorithm is used to process the identified broken areas, and by designing reasonable structure elements for dilation and erosion operations, the system intelligently connects the broken stripe segments and maintains the original stripe morphological features, generating preliminary repair image data.

[0071] In order to accurately obtain the geometric features of the repaired stripe, the system uses the Canny edge detection algorithm to process the preliminary repair data. The Canny algorithm extracts a precise and coherent set of stripe edge feature points through steps such as Gaussian filtering, gradient amplitude calculation, non-maximum suppression, and double-threshold detection.

[0072] In view of the linear characteristics of the structured light stripes, the system further uses Hough transform to analyze the edge feature point set. Hough transform can detect the straight line components in the image and connect the discrete edge points into complete straight line features to generate a linear distribution feature set describing the direction, position and length of the stripes.

[0073] The system analyzes the distribution characteristics of the broken area from the linear distribution feature set, and pays special attention to the density distribution of the broken points. The median filter technique is used to smooth the density distribution data, effectively eliminating abnormal values and discontinuous points, and obtaining smoother density data that can better reflect the true broken pattern.

[0074] Based on the smoothed density data, the system uses K-nearest neighbor algorithm to classify the feature points of the broken area. The algorithm classifies the feature points into different categories according to their similarity in spatial distribution, density characteristics and geometric properties, and generates a classified feature point set.

[0075] Finally, the system matches the classified feature point set with the pre-established stripe template database. By calculating the feature similarity and spatial distribution consistency, the most suitable repair template is selected, and the corresponding repair strategy is applied, finally outputting high-quality structured light stripe repair data, providing accurate and reliable input for subsequent three-dimensional measurement.

[0076] S108: Calculate the deviation amount according to the repair data combined with the principle of triangulation, process the deviation amount using an outlier rejection algorithm to obtain deviation data; The deviation amount is calculated according to the repair data combined with the principle of triangulation, and the deviation amount is processed using an outlier rejection algorithm to obtain deviation data, including: Extract the position and spatial information from the repair data, calculate the initial deviation amount value using the triangulation method, and generate the initial deviation amount value data; For the initial deviation amount value data, use the Isolation Forest algorithm to detect outliers, determine the outlier points by calculating the anomaly score, and generate the deviation data after removing the outlier points; Obtain the spatial distribution characteristics of the deviation amount value from the deviation data, and use the mean filter to smooth the distribution data to obtain the smoothed deviation distribution data; According to the smoothed deviation distribution data, use the DBSCAN algorithm to cluster the feature points of the deviation area to generate a clustered feature point set; Match the clustered feature point set with the pre-established distribution template database, calculate the distance deviation of each feature point from the template using the Euclidean distance, and determine the optimized deviation distribution data; Obtain the local density data from the optimized deviation distribution data, and smooth the density data using the median filter to obtain the final deviation amount value data.

[0077] In step S108, the system accurately calculates the misalignment amount of the pipe group pair based on the repaired high-quality stripe data combined with the optical triangulation principle, and ensures the accuracy and reliability of the final measurement results through advanced outlier processing technology.

[0078] The specific implementation process is as follows: First, the system accurately extracts the position and spatial information of the stripes from the repaired data. Using the structured light triangulation principle, the three-dimensional coordinates of each measurement point are calculated by calculating the geometric relationship between the projector, the camera and the measured surface, and then the initial deviation value data representing the misalignment of the pipe are obtained.

[0079] Considering that there may be residual noise or local interference in the actual measurement environment, the system uses the Isolation Forest algorithm to detect outliers in the initial deviation value data. This algorithm calculates the anomaly score of each data point by randomly partitioning the data space, and can efficiently identify and eliminate outliers that deviate from the main distribution, generating purified deviation data.

[0080] In order to further improve the distribution characteristics of the data, the system analyzes the spatial distribution pattern of the deviation data. The mean filter is used to smooth the spatial distribution data of the deviation amount, effectively suppressing random fluctuations while retaining the true deviation trend, obtaining smoothed deviation distribution data.

[0081] Based on the smoothed distribution data, the system uses the DBSCAN clustering algorithm to intelligently group the feature points in the deviation region. This algorithm is based on the concept of density reachability, and can automatically identify point sets with similar deviation characteristics and aggregate them into different cluster groups, generating a set of clustering feature points.

[0082] The system matches the clustered feature point set with the pre-established typical deviation distribution template database. By calculating the Euclidean distance between each feature point and the template, the similarity is quantified, and the optimal deviation distribution pattern is determined, obtaining optimized deviation distribution data.

[0083] Finally, the system extracts local density features from the optimized deviation distribution data and uses a median filter to smooth the density data. The median filter can effectively eliminate impulse noise in the density data while preserving the edge characteristics of the density distribution, and finally outputs high-precision final deviation value data, providing reliable quantitative basis for pipe group quality evaluation.

[0084] S109: Obtain depth map data, process the depth map data through classification segmentation, adjust the calculation range in combination with the deviation data and the quantization data, and generate quantization data.

[0085] In step S109, the system performs final data fusion and accurate calculation.

[0086] Firstly, the system acquires the raw depth map data collected by the three-dimensional sensor. This data contains the depth information of each pixel point in the measured area, forming the basic point cloud describing the three-dimensional topography of the surface.

[0087] Subsequently, the system classifies and segments the depth map data. This process usually uses a semantic segmentation algorithm based on machine learning or deep learning, aiming to classify each point or region in the depth map into different semantic categories, such as bevel slope, blunt edge plane, pipe base material surface, and possible weld spatter or rust area. Through this step, the system accurately identifies and isolates the bevel target area that needs to be analyzed, and defines the range for subsequent accurate calculation.

[0088] Next, the system starts a key data fusion and adaptive adjustment process. It uses the deviation data (such as the misalignment calculated in S108) and quantitative data (such as the bevel angle and blunt edge thickness calculated in S105) as important prior information and constraints.

[0089] Combined with deviation data adjustment: the system uses misalignment information to correct the relative position of the calculation area. For example, if the misalignment of the upper and lower pipe openings is detected, the system will perform coordinate translation or rotation alignment on the bevel calculation areas of the two pipe openings accordingly, ensuring that subsequent size measurement is performed in the correct spatial relationship, thereby eliminating the coupling effect of misalignment on bevel size measurement.

[0090] Combined with quantitative data adjustment: the system uses the preliminary calculated geometric parameters (such as the estimated range of angle and thickness) to define a more accurate and reasonable calculation range. For example, the angle search range is constrained within ±5° of the initial value, and the measurement area of the blunt edge thickness is accurately limited to the classified blunt edge area, avoiding calculation in irrelevant areas, thereby effectively improving calculation efficiency and anti-interference ability.

[0091] After the above calculation range dynamic adjustment based on prior information, the system performs final fine quantification calculation on the target areas (such as blunt edge and bevel face) classified and segmented in the optimized accurate area. This includes fitting the bevel slope plane equation using the least squares method to calculate the accurate angle, and accurately calculating the thickness distribution of the blunt edge area in the normal direction.

[0092] Finally, the system outputs a set of final quantification data which has been verified and optimized by multiple data. This set of data not only contains the groove angle, root face thickness and other dimensions, but also may contain comprehensive parameters related to the group state. The reliability and accuracy of the data are significantly improved due to the data fusion and range optimization in the whole process, which provides the final basis for the welding quality judgment.

[0093] The processor 201 can be a central processing unit (CPU), or other general purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, etc. The general purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0094] The memory 202 can be an internal storage unit of the electronic device 200, for example, a hard disk or a memory of the electronic device 200. The memory 202 can also be an external storage device of the electronic device 200, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 200. Further, the memory 202 can include both the internal storage unit and the external storage device of the electronic device 200. The memory 202 is used to store the computer program 203 and other programs and data required by the electronic device. The memory 202 can also be used to temporarily store the data that has been output or will be output.

[0095] In the embodiments provided by the present disclosure, it should be understood that the disclosed apparatus / electronic device and method can be implemented in other manners. For example, the described apparatus / electronic device embodiments are merely schematic. For example, the division of the modules or units is merely logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0096] The integrated modules / units, if implemented in the form of software functional units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program can include computer program code, which can be in the form of source code, object code, executable files or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read only memory (Read Only Memory, ROM), random access memory (Random Access Memory, RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electric carrier signal and telecommunication signal.

[0097] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure.

Claims

1. A method for measuring weld parameters against interference, characterized in that, include: The scattered light distribution data of a highly reflective surface is obtained by laser scattering technology, and the scattered light distribution data is subjected to Fourier transform processing to obtain spectral characteristics; Based on the spectral characteristics, correction coefficients are matched in a pre-established model, and the scattered light distribution data are weighted and adjusted using the correction coefficients to generate a corrected signal. The initial measured values ​​of the geometric parameters are extracted from the corrected signal, and the initial measured values ​​are adjusted through smoothing to obtain the geometric data; The boundary feature signal is separated from the corrected signal, and the boundary feature signal is processed by an edge enhancement algorithm to generate boundary enhancement data; A three-dimensional contour model is constructed using the boundary enhancement data and the geometric data. The three-dimensional contour model is then quantized using segmentation techniques to obtain quantized data. The original image of the projection data is acquired, and the original image is denoised by filtering to generate a filtered image; Continuous features are extracted from the filtered image, and the continuous features are processed to generate repaired data; The deviation is calculated based on the repaired data and the principle of triangulation. The deviation is then processed using an outlier removal algorithm to obtain the deviation data. Depth map data is acquired, processed by classification and segmentation, and the calculation range is adjusted by combining the deviation data and the quantization data to generate quantization data.

2. The method according to claim 1, characterized in that, The process involves acquiring scattered light distribution data from a highly reflective surface using laser scattering technology, performing Fourier transform on the scattered light distribution data to obtain spectral characteristics, including: Initial light distribution data of highly reflective surfaces are collected using laser scattering technology, and the distribution pattern of scattered light at different angles is recorded to obtain an initial light distribution dataset. Scattered light intensity information is extracted from the initial light distribution dataset, and the light intensity value at each angle is normalized to obtain scattered light intensity sequence data. The scattered light intensity sequence data is processed using a fast Fourier transform algorithm to generate a spectral data set; Spectral peaks and frequency distribution features are extracted from the spectral data set, and significant feature points are determined based on a preset threshold to obtain a spectral feature set; The spectral feature set is reduced in dimensionality using principal component analysis algorithm to obtain a dimensionality-reduced feature vector set. The reduced feature vector set is trained using a support vector machine classifier to generate a classification model and obtain surface characteristic determination data.

3. The method according to claim 1, characterized in that, The step of matching correction coefficients in a pre-established model based on the spectral characteristics, and then using the correction coefficients to weight and adjust the scattered light distribution data to generate a corrected signal includes: The statistical feature values ​​of the distribution data are obtained through the spectral features, and the statistical feature set is obtained by using the mean and variance calculation methods. The amplitude information corresponding to significant frequencies is extracted from the statistical feature set, and the amplitude significance is filtered according to a preset threshold to obtain the amplitude feature set; The deviation value is calculated based on the amplitude feature set and the standard amplitude data in the pre-established model. The combination of correction coefficients is determined by the mapping relationship between the deviation value and the correction coefficient. The scattered light distribution data is weighted and adjusted using the combination of correction coefficients to obtain a corrected signal dataset; The peak offset of the frequency distribution is extracted from the corrected signal dataset, and the offset distribution pattern is determined by a support vector machine classifier to obtain the offset classification result; The offset classification results are reduced in dimensionality using principal component analysis to obtain a set of feature vectors.

4. The method according to claim 1, characterized in that, The initial measured values ​​of geometric parameters extracted from the corrected signal are then adjusted through smoothing processing to obtain geometric data, including: The spectral distribution characteristics of the geometric parameters are obtained through the modified signal, and the spectral distribution is decomposed by Fast Fourier Transform to obtain the spectral feature set. Significant frequency components related to geometric parameters are extracted from the spectral feature set, and frequency significance is filtered according to a preset threshold to obtain the frequency feature set; The frequency deviation value is calculated based on the frequency feature set and the standard frequency data in the preset constraints. The adjustment coefficient is determined by using the mapping relationship between the deviation value and the geometric parameters. The initial measurements are weighted and adjusted using the adjustment coefficients to obtain geometric data. If the deviation between the geometric data and the preset constraints exceeds a preset threshold, the geometric data is smoothed using a Gaussian filtering method to obtain the new geometric data. Statistical features of the distribution are extracted from the geometric data, and a set of statistical features is obtained by using the mean and standard deviation calculation method.

5. The method according to claim 1, characterized in that, The step of separating the boundary feature signal from the corrected signal and processing the boundary feature signal using an edge enhancement algorithm to generate boundary enhancement data includes: The boundary feature signal is separated by signal decomposition using the corrected signal, and then frequency domain analysis is performed on the boundary feature signal using wavelet transform technology to obtain a frequency domain feature set. Significant boundary frequency components are extracted from the frequency domain feature set, and frequency significance is filtered according to a preset threshold to obtain the boundary frequency set; The boundary frequency set is convolved using a two-dimensional Gaussian kernel to obtain the boundary enhancement set; Based on the boundary enhancement set and the boundary feature signal, principal component analysis is used to reduce the dimensionality of the enhanced data to obtain feature-reduced data. If the deviation between the reduced feature data and the preset boundary constraints exceeds a preset threshold, the reduced feature data is smoothed using a Gaussian filtering method to obtain a smoothed feature set. Boundary distribution statistical features are extracted from the smoothed feature set, and the mean and variance are calculated to obtain the statistical feature set.

6. The method according to claim 1, characterized in that, The process involves constructing a three-dimensional contour model using the boundary enhancement data and the geometric data, and then using segmentation techniques to quantize the three-dimensional contour model to obtain quantized data, including: Using the boundary enhancement data and the geometric data, the target region is spatially discretized using triangulation mesh generation technology to generate a discrete mesh set; Curvature values ​​of each grid vertex are extracted from the discrete grid set to form surface curvature features; Gaussian filtering is used to smooth the curvature features to obtain a curvature feature set. For the curvature feature set, principal component analysis is used to compress the feature dimension to three dimensions, generating a dimensionality-reduced feature set; If the standard deviation of the dimensionality-reduced feature set exceeds a preset threshold, the dimensionality-reduced feature set is smoothed by mean filtering to obtain a smoothed feature set. Based on the smoothed feature set, the feature points are grouped using the K-means clustering algorithm to generate grouped feature sets.

7. The method according to claim 1, characterized in that, The step of extracting continuous features from the filtered image and processing the continuous features using a repair algorithm to generate repaired data includes: Continuity features are extracted from the filtered image, and continuous quantization data is obtained by calculating the ratio of continuous length to spacing. If the continuity quantization data is lower than a preset threshold, a morphological reconstruction algorithm is used to perform expansion and erosion operations on the fractured area to generate preliminary repair data. Using the preliminary repaired data, the boundary point set of features is extracted using the Canny edge detection algorithm to obtain boundary feature data; For the boundary feature data, the linear distribution characteristics are detected by Hough linear transform to generate a linear distribution feature set; The distribution density of the fracture region is extracted from the linear distribution feature set, and the distribution density is smoothed by median filtering to obtain smoothed density data; Based on the smooth density data, the K-nearest neighbor algorithm is used to classify the feature points of the fracture region and generate a set of classified feature points. Repair data is obtained by matching the set of classification feature points with a pre-established template database.

8. The method according to claim 1, characterized in that, The step involves calculating the deviation based on the repaired data and the triangulation principle, and then processing the deviation using an outlier removal algorithm to obtain deviation data, including: The location and spatial information are extracted from the repaired data, and the initial deviation value is calculated using triangulation to generate the initial deviation value data. For the initial deviation data, the isolated forest algorithm is used to detect outliers, outlier points are determined by calculating outlier scores, and deviation data is generated after removing outliers. The spatial distribution characteristics of the deviation values ​​are obtained from the deviation data, and the distribution data is smoothed using a mean filter to obtain smoothed deviation distribution data. Based on the smoothed deviation distribution data, the DBSCAN algorithm is used to cluster the feature points in the deviation region to generate a clustered feature point set. The clustered feature point set is matched with a pre-established distribution template database, and the distance deviation between each feature point and the template is calculated using Euclidean distance to determine the optimized deviation distribution data. Local density data is obtained from the optimized deviation distribution data, and the density data is smoothed by a median filter to obtain the final deviation value data.

9. The method according to claim 1, characterized in that, After generating quantized data, the method further includes: The quantified data is compared with the preset bevel size standard range. If it exceeds the standard range, an alarm signal is generated and adjustment suggestions are output.

10. An electronic device, characterized in that, include: One or more processors; A storage unit for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the method according to any one of claims 1 to 9.

Images