187 results about "Fitting algorithm" patented technology

The invention discloses an automobile seat framework machining defect detection method based on machine vision, and particularly relates to the technical field of defect detection. The method comprises the following steps: constructing a multi-angle image acquisition and edge reflection modeling module aiming at complex defects such as weld joint pseudo soldering, microcracks, hole site deviation and collapse deformation, extracting weld joint continuity, edge integrity and hole site geometric consistency characteristics by using a deep neural network, and generating a structural feature vector; defect type recognition and credibility scoring are completed through small sample anomaly modeling and Gaussian mixture model classification, sub-pixel-level coordinate labeling of defect positions is achieved in combination with a Gaussian fitting algorithm, a defect positioning map is output, traceability analysis and severity grading are conducted based on historical data comparison, and the defect positioning accuracy is improved. The method is suitable for industrial online detection and quality closed-loop control.

The invention discloses a Beidou and UWB dual-mode cooperative rail transit positioning method, and particularly relates to the technical field of fusion positioning, and the method comprises the following steps: constructing a cross-modal distortion discrimination array based on the space coverage state of two types of signals in a train path, recognizing a track mutation region, and dynamically adjusting the signal fusion weight; abrupt change points are corrected through a constraint backtracking trajectory fitting algorithm, and pseudo trajectory points are eliminated in combination with a positioning base station space topological structure; performing time sequence analysis on the signal fusion weight change, and if a frequent signal hopping trend exists, lowering the related fusion weight to the lowest tolerance threshold value to improve the track continuity and the positioning robustness; according to the method, the cross-modal distortion discrimination array is constructed, accurate identification of a track mutation region is realized, abnormal track points are effectively corrected in combination with fusion weight dynamic adjustment and a constraint backtracking track fitting algorithm, a signal jump trend is suppressed through a time sequence analysis mechanism, and the stability and positioning precision of track fusion are improved.

The invention belongs to the technical field of semiconductor device performance testing, and particularly relates to a VCSEL epitaxial wafer resistivity distribution measuring method and system, and the method comprises the steps: constructing a three-dimensional discretization model representing an epitaxial wafer and an unknown parameter vector containing the vertical resistivity, the top transverse resistivity and the substrate leakage junction nonlinear resistance; a physical response function is defined, the vector is iterated through a discrete point fitting algorithm, so that the total residual error between the analog voltage difference output by the function and the actually measured voltage value under the cross-order-of-magnitude current is minimum, and the real vertical resistivity is extracted from the optimal parameter vector; and obtaining the vertical resistivity of each point on the two-dimensional test point grid so as to generate a distribution diagram and carry out anomaly detection. According to the invention, the problem of measurement distortion caused by the leakage defect of the substrate is solved, and the measurement accuracy of the vertical resistivity is improved.

The invention discloses a ship water gauge scale line fitting algorithm, and relates to the technical field of ship water gauge scale calculation, and the method comprises the steps: obtaining a ship water gauge region image through an image collection device to extract a measurement point coordinate, and building a mapping relation between a pixel position and a physical draft; a weighted least square method is adopted to perform nonlinear curve fitting, a weight coefficient is calculated based on a local neighborhood standard deviation, a polynomial order p is dynamically adjusted according to an adjusted decision coefficient, and the decision coefficient is iteratively optimized to improve the balance between amplitude and model complexity. And the system repeatedly corrects model parameters through a dynamic verification mechanism of the residual sum of squares and a preset threshold value, and finally an accurate water gauge scale fitting curve is generated. The method effectively solves the problems that traditional manual observation is prone to environmental interference and large in measurement error, high robustness is kept in a complex water area environment, and reliable data support can be provided for the fields of ship load evaluation, channel safety management and the like.

The invention relates to the technical field of semiconductor detection, in particular to a method and a system for measuring the thickness of a semiconductor epitaxial layer based on multi-beam interference correction. Acquiring infrared reflection spectrum data of the wafer under different incident angles, and performing double-stage smoothing preprocessing on the data by using a moving least square fitting algorithm; an interference extreme point is accurately identified by adopting a self-adaptive extreme value feature extraction strategy; the method comprises the following steps of: calculating the dynamic refractive index and the initial thickness estimation value of the epitaxial layer, constructing a multi-beam interference intensity distribution model, and finally providing a hybrid nonlinear optimization algorithm to invert the accurate thickness d of the epitaxial layer, thereby overcoming the measurement error caused by neglecting the multi-beam effect and the refractive index change in the traditional double-beam interference, and improving the accuracy of the multi-beam interference. And the precision and the reliability of epitaxial layer thickness measurement are remarkably improved through the combination of data preprocessing and a nonlinear optimization algorithm. The objective of the invention is to solve the problem of how to correct measurement errors caused by multi-beam effect and refractive index change.

The invention relates to the technical field of textile equipment testing, in particular to a yarn textile high-speed spindle operation stability testing method and system, and the method comprises the following steps: S1, testing element layout; s2, initializing the system; s3, carrying out transient test; s4, testing a steady state; s5, interference processing: carrying out spectrum analysis on the originally acquired data, identifying interference frequency characteristics, eliminating interference by adopting a notch filtering and Kalman filtering combined method, and verifying a filtering effect; s6, parameter correlation modeling: constructing a vibration-temperature-tension-rotating speed parameter correlation model by adopting a control variable method, solving model parameters through a fitting algorithm and verifying validity; s7, quality tracing; s8, stability judgment; and S9, performing real-time monitoring and regulation. According to the method, a combined filtering scheme of notch filtering and Kalman filtering is adopted, fixed interference and random interference such as power frequency and mechanical resonance are suppressed in a targeted manner, and through comparison of locally amplified signals, the filtered signals are smooth in waveform and thorough in noise removal, and the signal-to-noise ratio and reliability of vibration detection signals are improved.

The invention provides a laser radar attitude compensation wind measurement method and system, and relates to the technical field of laser radar wind measurement, and the method comprises the steps: obtaining the six-degree-of-freedom motion data of a carrier; driving the two-axis stabilized platform to compensate based on the three-axis rotation angle so as to stabilize the direction of the laser beam, and scanning to obtain initial radial wind speed data; then synchronizing the translation speed with the wind speed data by using a linear fitting algorithm to obtain a synchronous data set; a laser radar scanning mode is determined, and a corresponding attitude compensation strategy is selected according to the mode; inputting the synchronous data set and the strategy into a pre-training machine learning model to generate a compensation parameter, and processing the initial data through Kalman filtering based on the parameter to obtain target radial wind speed data; and finally, performing wind field inversion calculation based on the target data to generate a wind field measurement result. According to the invention, the wind field measurement precision of the laser radar on the motion platform under the complex sea condition is improved.

The invention relates to the technical field of flow-state solidified soil detection, in particular to a flow-state solidified soil online flow value detection and feedback method and system, and the method comprises the steps: collecting multi-source signal data of flow-state solidified soil; adopting a data fitting algorithm to judge an equivalent flow value of the flow-state solidified soil; constructing a flow value prediction model of the flow-state solidified soil, and taking the equivalent flow value as the input of the flow value prediction model to obtain a real-time flow value of the flow-state solidified soil; comparing the real-time flow value with a preset target flow value, and automatically adjusting production control parameters of the flow-state solidified soil according to a comparison result; and visualizing the real-time flow value curve of the flow state solidified soil. Through the system and the method, the problems of lagging, poor precision and lack of a feedback mechanism in flow-state solidified soil flow value detection are effectively solved, online flow value detection in the flow-state solidified soil production process is realized, manual sampling is not needed, flow speed, pressure and visual signals are fused in flow value detection, the precision is high, and the response is fast.

The invention relates to the technical field of signal processing, in particular to a multi-dimensional force signal processing method and system, and the method comprises the steps: obtaining analog signals outputted by a plurality of strain bridges, carrying out the amplification and filtering of the analog signals, and converting the analog signals into digital signals; under the standard force load, the relation between the digital signal and the corresponding load is obtained, and a calibration data set is obtained; establishing a mapping model between the digital signal and the multi-dimensional force based on the calibration data, and determining decoupling parameters by adopting a fitting algorithm; the digital signals collected in real time are calculated based on the decoupling parameters, and force values of all dimensions are obtained; and outputting the force value of each dimension through a plurality of communication interfaces, and carrying out online calibration on the decoupling parameters. According to the invention, the problem of insufficient precision caused by coupling interference among different channels in a multi-dimensional force signal processing process in the prior art is effectively solved, and high-precision and stable multi-dimensional force value output is realized.

The invention discloses a machining method suitable for a large marine propeller, and relates to the technical field of precise combined machining technologies of the large marine propeller. The machining method comprises the steps that the propeller and a standard ball are installed on a machine tool rotating table, a high-precision calibration needle is installed on a main shaft, and a fitting algorithm is used for establishing a positioning reference of a workpiece; constructing a workpiece coordinate system and a conversion relation between the workpiece coordinate system and a machine tool coordinate system; acquiring point cloud data of each part of the propeller, and unifying the point cloud data to a workpiece coordinate system through coordinate conversion; performing coarse and fine splicing on the blade root point cloud, and forming workpiece coordinate system point cloud data of the propeller in combination with data of other parts; a normal projection algorithm is adopted to calculate the polishing amount based on the data; and according to the polishing amount, generating a machining track containing blade thickness calibration, and according to the machining track, milling and polishing the propeller. According to the method, the positioning precision is remarkably improved, the clamping error is effectively eliminated, the grinding amount is accurately and efficiently calculated, the adaptability is high, the machining efficiency is remarkably improved, and the production period is shortened.

The application discloses a macro modeling method for ensuring direct current precision, and the method comprises the following steps: reading n-port network parameters; initializing sampling points for constructing a macro model; initializing the order of the macro model; initializing the pole of the macro model; establishing a weighted linear system according to the current order, the sampling points and the pole; solving a linear equation set, judging convergence according to an error, and obtaining the macro model; and outputting the optimized macro model by applying a residue matrix disturbance to optimize the macro model. The macro modeling method for ensuring direct current precision provided by the application controls the direct current precision in two stages. In the first stage, a weighted least square method is adopted on the basis of a linear system established by a traditional vector fitting algorithm to preliminarily control the direct current precision. In the second stage, a small disturbance is applied to the residue matrix of the macro model, so that the disturbed macro model has accurate direct current characteristics.

The invention discloses a self-adaptive variable-diameter spiral groove numerical control machining method and device based on wood curvature perception, and belongs to the technical field of wood spiral groove machining. The method comprises the following steps that S1, wood feeding and positioning are conducted, square timber to be machined is fixed through a self-adaptive clamp, and it is ensured that the axis of the square timber is aligned with the machining standard; s2, real-time data collection is conducted, specifically, the front face, the back face and the two side faces of the batten are scanned through multiple sets of laser displacement sensors, the curvature and moisture content data of the wood are collected in a high-precision mode through the multiple sets of laser displacement sensors, and a customized tool path matched with irregular deformation of the wood is generated in combination with a least square method curvature fitting algorithm; and meanwhile, a cutting pressure dynamic compensation mechanism is matched, the cutting pressure is stabilized within a reasonable range, heat expansion and cold contraction deformation of the groove profile is avoided through composite cooling shaping after machining, finally, the groove profile machining precision reaches + / -0.05 mm, the problems that traditional fixed parameter machining is poor in adaptability and low in size consistency are thoroughly solved, and the follow-up assembly difficulty is greatly reduced.

The invention relates to the technical field of intelligent driving, in particular to a sensing element and map data fused lane modeling method, system and device and a medium, and the method comprises the steps: obtaining sensing point cloud data of lane elements around a vehicle and map point cloud data around the current position of the vehicle in real time; performing coordinate system conversion and time synchronization on the sensing point cloud data and the map point cloud data; performing registration by using an iteration nearest point, and calculating a registration confidence coefficient; calculating a perception confidence coefficient and an initial map confidence coefficient, correcting the initial map confidence coefficient based on the registration confidence coefficient to obtain a corrected map confidence coefficient, and distributing a fusion weight based on the perception confidence coefficient and the corrected map confidence coefficient; and based on the fusion weight, fitting a parameterized curve through a weighted fitting algorithm, and generating a fusion lane model. According to the method, adaptive optimization fusion of the perception data and the map data can be realized, and the precision, stability and environmental adaptability of lane modeling are improved.

The present application belongs to the technical field of semiconductor device performance testing, and particularly relates to a VCSEL epitaxial wafer resistivity distribution measurement method and system, which comprises the following steps: constructing a three-dimensional discretization model representing the epitaxial wafer and an unknown parameter vector containing a vertical resistivity, a top lateral resistivity and a nonlinear resistance of a substrate leakage junction; defining a physical response function, and through a discrete point fitting algorithm, iterating the vector to minimize the total residual between the simulated voltage difference output by the function and the measured voltage value under a cross order of magnitude current, and extracting the real vertical resistivity from the optimal parameter vector; and obtaining the vertical resistivity of each point on a two-dimensional test point grid to generate a distribution map and perform anomaly detection. The present application solves the measurement distortion problem caused by substrate leakage defects and improves the measurement accuracy of the vertical resistivity.

The invention discloses a mechanical arm control method and system for high-precision machining, and relates to the technical field of machining control. According to the method, an operation data set is constructed, a lead screw displacement error Wx (t) is calculated, and a mapping relation between time and an angle is established through a lead screw transmission ratio to obtain an angle error vector epsilon (phi); superposing and averaging the angle error vector by adopting a periodic signal superposition technology to obtain a lead screw angle error amount Es (phi); solving a fitting parameter group through a least square fitting algorithm, constructing a screw pitch error index Wwc (phi), performing dynamic error compensation on a screw rotation angle variable phi, and generating a corrected path track Xc; and calculating a periodic residual error ratio Rre of an error before and after correction by extracting actual displacement and theoretical displacement data, and evaluating a compensation state. According to the method, through time and angle domain collaborative modeling and periodic residual error closed-loop evaluation, dynamic correction and precise control over the motion error of the mechanical arm lead screw are achieved.

The application discloses a kind of based on multi-focal sharpness analysis's stacked cell hierarchy determination method and system.Acquire Z-stack multi-focal image sequence and extract each focal plane cell contour;Based on centroid coordinates and area features, establish the profile correlation data table of the same cell in different focal planes;Ellipse fitting algorithm is used to complete the shape of the profile of the cell that is blocked;Based on complete geometry mask, screen potential stacked candidate cell pair and locate overlapping area profile;With gradient function as evaluation index, calculate the overlapping area profile sharpness of each cell under multiple focal planes, determine the best focus focal length and judge cell hierarchy relationship;Attribute graph is constructed and topological sorting is executed, and global stacked hierarchy sequence is output.The application establishes hierarchy reasoning rule with rigorous physical basis according to the physical law that upper cell best focus focal length is higher, is compatible with existing Z-stack imaging process, adapts to basic laboratory high-throughput analysis demand.

The invention discloses a molded surface error analysis method and system based on an on-machine measurement detection result, and the method comprises the steps: S1, carrying out the on-machine measurement of the surface of a machined workpiece, and obtaining an actual measurement point set; s2, fitting the actual surface of the workpiece through a fitting algorithm based on the actual measurement point set; s3, acquiring a theoretical surface of the workpiece, and registering the theoretical surface and an actual surface by adopting an iterative nearest point algorithm; s4, based on the theoretical surface, acquiring a theoretical point set by adopting a self-adaptive discrete mode; s5, projecting each theoretical point in the theoretical point set to an actual surface to obtain a corresponding actual point, and calculating a normal error between each theoretical point and the actual point; and S6, based on the normal error, generating an error cloud picture for visualization. The method has the advantages of high profile error analysis precision and the like.

The invention discloses a load transfer line prediction method and system based on LSTM, and belongs to the technical field of power system automation, and the method comprises the steps: S1, obtaining a first prediction load through a historical load of a to-be-transferred line and a load prediction model; s2, fitting the first predicted load through a fitting algorithm to obtain a first fitting curve, and obtaining the importance degree of the to-be-transferred line according to the first fitting curve; s3, obtaining a second predicted load through the historical load of the to-be-received line, and fitting the second predicted load through a fitting algorithm to obtain a second fitting curve; s4, whether the first fitting curve and the second fitting curve meet preset conditions or not is judged, if yes, a load transfer line is obtained, and if not, a necessary to-be-transferred line is obtained through the importance degree, and the load transfer line is obtained according to the first predicted load and the second predicted load corresponding to the necessary to-be-transferred line. The technical problems that the utilization rate of power grid resources is difficult to improve and the power supply stability of important power utilization scenes is difficult to guarantee in the prior art are solved.

The application provides a pulse eddy current pipe wall thickness detection method based on an RSBO optimized LSTM network, pulse eddy current signals are used to estimate the thickness of the pipe, a data sample length evaluation criterion based on a least square fitting algorithm is used to determine the interval of the retained measurement signals, the processed signals are input into an LSTM model to detect the thickness of the pipe, and a combination optimization based on random search and Bayesian optimization is used to set the hyperparameters of the LSTM model.

This application discloses a temperature curve fitting algorithm, including the following steps: (1) preheating temperature control and feature extraction; (2) constant power curve fitting; (3) volume estimation; (4) anomaly handling. The temperature curve fitting algorithm of this application stabilizes to 50℃±3℃ within 10 minutes, and mines the temperature data features of the preheating stage through sliding residual trend analysis, capturing the weak heat capacity influence of small-volume insects in advance, and providing a sensitivity marker for the slope extraction in the subsequent constant power stage; the fitting mode is automatically switched by the slope reduction threshold, which not only retains the accurate extraction of the core slope in the stable segment, but also avoids the fitting distortion in the slowing segment; the fitting benchmark is dynamically corrected by h, rather than relying on a fixed insect-free curve, which effectively offsets the heat dissipation difference interference caused by room temperature fluctuations; when the heat dissipation coefficient estimation is abnormal, it is automatically downgraded to a linear fitting benchmark, and when historical data is insufficient, a backup formula is called to ensure that effective results can still be output under extreme scenarios such as sensor fluctuations and data loss.

The invention relates to the field of exploration, and particularly discloses a method and device for measuring the permeability coefficient of an in-situ soil body, computer equipment and a readable storage medium, and the method comprises the following steps: obtaining a plurality of soil samples with preset burial depths, and obtaining the initial void ratio of each soil sample; carrying out a consolidation test on the soil samples to obtain the compression amount, the total compression amount and the test value of the consolidation coefficient of the sampling point of each soil sample under a plurality of preset consolidation pressures per minute; based on the initial void ratio, the preset burial depth, the preset consolidation pressure, the compression amount per minute, the total compression amount and the consolidation coefficient test value of the sampling point of each soil sample, a permeability coefficient prediction model is obtained by combining a fitting algorithm; and obtaining the burial depth and the corresponding pressure of the to-be-predicted soil body, and calculating the permeability coefficient of the to-be-predicted soil body based on the permeability coefficient prediction model. According to the method provided by the invention, sampling is carried out in soil layers with different burial depths, so that the real state of the soil body in the in-situ stratum can be restored, and the reliability and accuracy of a prediction result are remarkably improved.

The invention relates to the technical field of automatic micro-assembly, in particular to a micro-part multi-interference complex image positioning method for automatic micro-assembly. Comprising the steps that S1, a template matching or target detection deep learning algorithm is used for an original part picture a collected by a visual system, and a picture block b containing complete to-be-detected features is obtained; s2, on the basis of the image block b, performing image preprocessing by using a super-resolution reconstruction network and a texture removal network to obtain an image block c with texture interference removed and fuzzy feature clearness; s3, on the basis of the image block c, obtaining to-be-detected feature edge information d by using a binarization and adaptive Canny algorithm; s4, the obtained edge information d is a point set containing part of interference information, interference points are removed through a contour moment algorithm, an RANSAC algorithm and an iterative least square fitting algorithm, and the part pose is obtained through a fitting line or circle.

The invention provides a novel target and background contrast ratio measuring method, and belongs to the field of precise instrument manufacturing. When an image gray value is used for measuring the contrast ratio of a target and a background, imaging parameters of a detector and parameters of an optical system need to be calibrated, so that brightness information of the target and the background is truly reflected. The method comprises the following steps: respectively carrying out contrast measurement calibration on standard target plates with different contrast values under different illuminations by using a brightness meter, measuring the contrast values of the standard target plates by using an image gray value, fitting the contrast measured by the image gray value with the contrast measured by the brightness value in software by using a mathematical fitting algorithm, and calculating the contrast value of the standard target plates. The purpose of measuring the target background contrast value by using the image gray value is achieved. And establishing relational mapping between brightness and a gray value, defining a rule of a sampling sample, selecting a target and a background, and finally calculating a contrast ratio between the target and the background.

The invention discloses a peak-valley segmentation multi-size self-adaptive peak searching method applied to collision light pulse radiation, relates to the field of light radiation measurement, and aims at the characteristics of peak structure size difference, asymmetric structure distribution and the like of collision light radiation signal data, based on a peak-valley segmentation region and in combination with a logarithmic normal distribution fitting algorithm. According to the method, adaptive extraction of characteristic quantities of peaks of different sizes is realized, and main existing areas of peak structures of different scales are adaptively traversed through screening of peaks and valleys, so that the calculation amount of signal data processing and peak influence area screening is reduced, the peak searching process efficiency of the peak structures of different sizes is improved, and the calculation cost and time are saved.

The invention belongs to the technical field of spectral analysis, and discloses a plant cell wall microfibril orientation polarization Raman spectrum measurement method, which comprises: 1, carrying out polarization Raman spectrum data acquisition on a sample to be measured; step 2, acquiring required data based on a data acquisition result of the polarization Raman spectrum; 3, performing nonlinear parameter fitting on the required data, obtaining an optimal parameter combination and a corresponding coefficient through a minimum error function, and performing forced symmetry post-processing on the optimal Raman tensor matrix; step 4, calculating a microfilament angle of a corresponding angle according to the symmetrical tensor parameter; in the fourth step, based on the symmetrical tensor matrix, the angle needed for rotating the tensor matrix to the main axis direction is calculated; and carrying out normalization processing on the calculated angle value. The technical problems that in the prior art, due to the fact that a nonlinear fitting algorithm is complex and prone to being affected by noise and experimental errors, the fitting convergence effect is poor, and the stability and reliability of a detection result are insufficient are solved.

The invention relates to a river levee dam break flood prediction method and device, and belongs to the technical field of flood prediction, and the method comprises the steps: predicting the future boundary data of a river channel through employing a historical rainfall-river channel boundary fitting algorithm according to the forecast rainfall of a weather bureau, and carrying out the prediction of the river levee dam break flood based on the future boundary data of the river channel and a one-dimensional hydrodynamic coupling mathematical model; predicting a future river section water level and a future river section flow; based on the future river section water level and the future river section flow, determining the water depth and the flow velocity of a coupling interface by combining a discontinuous interface algorithm; the water depth and the flow velocity of the coupling interface are used as boundary conditions to act on the two-dimensional hydrodynamic coupling mathematical model, and a momentum flux correction method is combined to predict the future ground flood level, so that the problems of unbalanced still water and unstable prediction during flood level prediction are solved, and the flood level prediction accuracy is improved.

The application relates to the technical field of high-dimensional data, in particular to a high-dimensional data fitting algorithm and system based on an improved gradient descent, the algorithm is deeply coupled with a long short-term memory (LSTM) network, a gated differential adaptive momentum variable gradient descent (AM-VGD) is designed for high-dimensional time series data, high-precision fitting of high-dimensional data is realized, the gated differential adaptive momentum variable gradient descent (AM-VGD) is proposed, different momentum factors and gradient decoupling items are designed according to the gradient characteristics of the LSTM forget gate, input gate and output gate, the gradient propagation efficiency is improved by 40% under high dimension, the model convergence iteration number is reduced to <=500 times, the convergence speed is improved by more than 2 times, a stacked auto-encoder (SAE) is fused to decouple and extract high-dimensional features, a high-dimensional decoupling regular term of the gradient descent is combined, the collinearity interference between features is eliminated, the fitting determination coefficient R2 of high-dimensional time series data is greater than or equal to 0.98, the mean square error (MSE) is reduced by more than 60%, and there is no overfitting / underfitting phenomenon.

The present application relates to the technical field of material testing, in particular to a spectral temperature brightness degradation prediction method of polyurethane functional material component. It constructs a weighted joint regression network prediction model for predicting the polyurethane functional material component; the spectral similarity, temperature similarity and brightness similarity of the polyurethane functional material component are predicted through the weighted joint regression network prediction model. A functional material temperature, brightness and spectral feature degradation characterization model based on the weighted joint regression network is proposed, and through the weighted joint regression of the temperature, brightness and spectral feature image detection results, compared with the traditional one regression fitting algorithm, the situation of falling into local minimum solution and overfitting can be avoided, so that the prediction result is closer to the actual situation.

The application provides a welding seam extraction method based on multi-dimensional information fusion, comprising the following steps: S1, collecting a Texture image and depth data of a welding seam; S2, inputting the Texture image into an image recognition model to obtain a mask image of the welding seam; S3, performing binaryzation processing on the mask image to extract pixel coordinates of the welding seam; S4, aligning the pixel coordinates with the depth data to generate an initial welding seam point cloud; S5, processing the initial welding seam point cloud by using a plane segmentation and fitting algorithm to extract characteristic points of the welding seam; and S6, obtaining an accurate welding seam point cloud by using a straight line fitting algorithm or a plane intersection fitting algorithm. By combining two-dimensional image segmentation and three-dimensional point cloud processing technology, the application solves the problem that a traditional method is difficult to simultaneously consider accurate three-dimensional information extraction and anti-background noise interference of the welding seam in a complex industrial scene, improves the precision and robustness of welding seam extraction, and guarantees the processing efficiency, thereby providing a high-reliability three-dimensional path planning basis for an automatic welding system.

The application discloses a kind of vehicle delay calculation method and device, it is related to traffic control technical field, main purpose is to improve the precision of calculating vehicle delay result.The main technical scheme of the present application is: obtaining the low-frequency trajectory data corresponding to a plurality of vehicles in intersection area respectively;Linear fitting reconstruction is carried out on the low-frequency trajectory data of each vehicle using a predetermined fitting algorithm, to obtain the high-frequency trajectory data corresponding to each vehicle in intersection area;Based on the high-frequency trajectory data corresponding to each vehicle, the initial delay result corresponding to each vehicle in the intersection area is calculated;The initial delay result corresponding to each vehicle is modified respectively, to obtain the modified delay result corresponding to each vehicle;Based on the modified delay result corresponding to the plurality of vehicles, the target delay result of the vehicle in the intersection area is calculated.The present application is used to calculate the vehicle delay of intersection.