29 results about "Iterative thresholding" patented technology

The invention discloses a mold quality detection method and system based on image processing, and relates to the technical field of mold quality detection.The method comprises the steps that a high-precision industrial camera is installed over a mold to obtain a mold surface gray level image, and graying and smoothing are conducted on the obtained image; constructing a distribution model according to the reference distance distribution condition of each pixel point in the smoothed image, calculating a period strength degree, preliminarily screening the pixel points according to the period strength degree, and distinguishing the possibility of texture pixel points and non-texture pixel points; by using the difference between normal texture periodicity and crack randomness, through calculating the periodicity intensity degree, screening texture pixel points and non-texture pixel points, screening suspected texture pixel points, calculating the periodicity similarity, screening out a texture region and then carrying out iterative threshold segmentation, the detection problem caused by mixing of cracks and normal textures is solved; and the detection accuracy is improved.

The application provides a serial batch processing scheduling method based on a hybrid particle swarm algorithm in a fuzzy environment, and relates to the field of production scheduling.The method comprises the following steps: encoding and analyzing workpieces and machines based on machine indexes; in the case of parallel machine scheduling, initializing algorithm parameters based on the hybrid particle swarm algorithm, initializing a population by a heuristic algorithm, analyzing fitness, and updating the speed and position of particles; determining the minimization of the maximum completion time as an objective, executing a variable neighborhood descent local search strategy, and updating the local optimum and the global optimum; and in the case that an iteration index exceeds an iteration threshold, determining the end of the algorithm and outputting a target result to represent the allocation and scheduling information of the workpieces and machines.The application considers fuzzy processing time, learning effect and deterioration effect, and considers the scheduling problems of single machines and non-single machines in a fuzzy environment based on actual production conditions, which is helpful for iron and steel plants to formulate production strategies and reasonably arrange the allocation and scheduling of workpieces and machines.

The invention discloses an industrial big data processing method and system based on a cloud platform, and relates to the technical field of data processing.The method comprises the steps that edge nodes are deployed, a preset first machine model is loaded, reasoning training is executed on target industrial data, nonlinear damage factors are generated, and a first candidate set is screened out; uploading the first candidate set to a cloud platform, executing iterative training on the first candidate set by adopting a preset second machine model, and screening out a second candidate set; if the number of iterations reaches a preset iteration threshold value, a gradient distribution task is triggered, and the cloud platform is controlled to distribute gradient difference values according to the optimal gradient path; guiding the first machine model through the second machine model, training the first machine model through the first candidate set and the second candidate set to obtain an updated first machine model, and deploying the updated first machine model to each edge node for identifying nonlinear damage next time; according to the invention, the efficiency and accuracy of industrial big data processing are improved.

The invention provides an optical tweezer generation method and device, electronic equipment and a storage medium, and the method comprises the steps: initializing an optical tweezer phase based on a designated optical tweezer phase and an optical tweezer position mask matrix, and initializing an optical tweezer amplitude based on a target optical tweezer amplitude diagram; and updating the phase of the spatial light modulator based on the phase of the optical tweezers and the amplitude of the optical tweezers, respectively updating the phase of the optical tweezers and the amplitude of the optical tweezers based on the phase of the spatial light modulator, and carrying out phase constraint on the updated phase of the optical tweezers based on the appointed phase of the optical tweezers and the position mask matrix of the optical tweezers. Updating the number of iterations and returning to update the phase of the spatial light modulator based on the new optical tweezer phase and the optical tweezer amplitude until the number of iterations reaches an iteration threshold value; and generating the optical tweezers based on the phase of the spatial light modulator. According to the method, the device, the electronic equipment and the storage medium provided by the invention, the global optimization capability is solved by utilizing the iterative algorithm, the phase of the optical tweezers is ensured to be controllable, and the risk of random jumping of the phase of the optical tweezers is avoided.

The invention provides a quantum classical hybrid calculation method and device. The method comprises the following steps: performing linearization operation on an objective function and constraint conditions in a QCBO model; setting a variable matrix and a linear mapping operator according to the objective function after linearization operation, the constraint condition and the corresponding co-orthogonality optimization problem; according to the current variable matrix and the linear mapping operator, calculating to obtain a residual vector, and calculating to obtain a secondary unconstrained binary optimization coefficient matrix; solving the current secondary unconstrained binary optimization problem to obtain a current solution vector; determining an iteration direction according to the current solution vector, updating the variable matrix by using the first step length, and taking the updated variable matrix as a current variable matrix; and when the number of iterations is greater than a preset iteration threshold value, performing rounding operation on the current variable matrix, and extracting a vector from the variable matrix after the rounding operation as a final solution. By applying the method, the problem of 0-1 binary optimization with nonlinear constraints can be effectively solved.

The invention provides a serial batch processing scheduling method based on a hybrid particle swarm algorithm in a fuzzy environment, and relates to the field of production scheduling, and the method comprises the steps: carrying out the coding analysis of a workpiece and a machine based on a machine index; under the condition that the scheduling type is parallel machine scheduling, based on a hybrid particle swarm algorithm, initializing algorithm parameters, initializing a population through a heuristic algorithm, analyzing fitness, and updating the speed and position of particles; determining the minimum maximum completion time as a target, executing a variable neighborhood descent local search strategy, and updating local optimum and global optimum; and under the condition that the iteration index exceeds an iteration threshold value, determining that the algorithm is ended, and outputting a target result to represent distribution scheduling information of the workpiece and the machine. According to the method, fuzzy processing time, a learning effect and a deterioration effect are considered, single-machine and non-single-machine scheduling problems in a fuzzy environment are considered based on actual production conditions, and an iron and steel plant is helped to formulate a production strategy and reasonably arrange distribution scheduling of workpieces and machines.

The application relates to an ISAR sparse imaging method, device and equipment based on a DA-ISTA network. An iterative threshold convergence algorithm (ISTA) is used to construct a deep neural network for ISAR sparse imaging. In the ISTA, the iteration process is tiled into a multi-layer network structure, and the linear sparse transformation in the ISTA algorithm is replaced by a nonlinear convolution operation in each layer of the network structure. In the DA-ISTA network, a large number of iterations are not required as in the ISTA algorithm. Only a small number of network structure layers are required to process ISAR sparse aperture echo data, and a high-quality ISAR imaging result can be obtained. The method improves the imaging quality and the calculation efficiency, and makes the DA-ISTA network interpretable.

The application discloses a kind of uniform missing linear frequency modulation signal fast iterative threshold deconvolution reconstruction method, solve the matrix singularity problem that sparse reconstruction algorithm appears in pseudo-inverse process in the reconstruction of uniform missing linear frequency modulation signal, the present application is based on fast iterative threshold algorithm, according to the pattern of uniform missing, deduce the point spread function due to signal time domain missing causes spectrum aliasing, then utilize point spread function for the spectrum of zero padding and dechirp processing after missing linear frequency modulation signal by fast iterative threshold algorithm carries out complex deconvolution, to recover the spectrum of signal in frequency domain, in turn reconstruct complete signal.The present application realizes complex deconvolution by fast iterative threshold algorithm, need not carry out matrix pseudo-inverse operation, to avoid matrix singularity problem.

This invention provides a test case generation method, apparatus, device, and medium based on a large language model, relating to the field of software analysis and testing technology. The method includes: generating prompt words based on the variable data flow graph corresponding to the function to be tested and the function itself, and inputting these prompt words into a fine-tuned large language model to obtain initial test cases; obtaining evaluation metrics corresponding to the initial test cases; if the evaluation metrics corresponding to the initial test cases do not meet the evaluation conditions, returning to the step of inputting prompt words into the fine-tuned large language model to obtain initial test cases corresponding to the function to be tested, until the number of iterations is greater than or equal to an iteration threshold; and determining the test cases corresponding to the function to be tested based on the evaluation metrics corresponding to historical test cases stored in a historical test case set, as well as the evaluation metrics corresponding to the initial test cases. The technical solution of this invention can improve the automation and accuracy of test case generation.

The application belongs to the technical field of image compressive sensing reconstruction, and discloses an image compressive sensing reconstruction method and system based on optimization algorithm expansion. In the sampling stage, a convolution sampling method is used instead of a traditional random matrix sampling. In the reconstruction stage, a generalized iterative threshold shrinkage algorithm is expanded into a deep network, and a jump information connection structure is designed in a reconstruction submodule R. Residual modules are used to connect the feature information before and after the modules, so that the inherent information loss in the deep expansion network is avoided. Furthermore, a double-scale denoising module is designed at the back end of the reconstruction submodule R, and different scale features are combined to denoise the image. The application not only realizes the application of the algorithm expansion method in the image compressive sensing, but also improves the reconstruction effect by using the jump connection structure and the double-scale denoising module. The application has higher accuracy and better robustness.

The invention relates to a compressed sensing-based remote sensing image full-process adaptive coding and decoding method, and relates to the technical field of space remote sensing imaging. The method comprises the following steps of: calculating an image content complexity self-adaptive selection coding mode by utilizing a maximum between-class difference method; extracting an image feature factor, establishing a significance model of a corresponding mode, and guiding the coding tree block to carry out adaptive form partitioning; detecting the sparsity of the coding block through an information density function to mine the sparsity of the image, and realizing the adaptive sampling rate distribution of the image; and adaptively selecting a reconstruction matrix and a sparse base according to the category of the coding block, setting an iteration threshold according to the known sparseness, and reconstructing the coding block by using an orthogonal matching pursuit algorithm. According to the method provided by the invention, the compression ratio does not need to be manually preset, the sampling mode can be selected according to the image content, the compression efficiency is adaptively set by detecting the sparseness, and the requirements of on-satellite rapid compression and ground high-fidelity reconstruction are met.

The invention discloses a man-machine collaborative scheduling method and system based on an improved genetic algorithm, and the method comprises the steps: initializing an order scheduling environment, processing the imported order data, machine data and material data, and obtaining an initial order scheduling scheme; carrying out iterative optimization processing on the initial order scheduling scheme by adopting an improved genetic algorithm, circularly executing fitness calculation, selection, crossover and mutation operations when an iteration threshold value is not reached to obtain an updated scheduling strategy, and outputting a machine decision scheduling scheme when the iteration threshold value is reached; submitting the machine decision scheduling scheme to an artificial decision interface for determination processing, and receiving an artificial determination instruction generated based on analysis of the machine decision scheduling scheme; and processing the machine decision scheduling scheme according to the manual judgment instruction, and selecting to re-execute iterative optimization of the improved genetic algorithm or output the machine decision scheduling scheme as a final scheduling scheme. According to the invention, efficient optimization of order scheduling and improvement of decision accuracy can be realized.

The application discloses a kind of defect detection method and system of PHEMT, it is related to electronic device detection technical field, method includes: obtaining the surface image of PHEMT;The surface image is preprocessed, and target image is determined;The gradient response image of the target image is calculated;The gradient response image is iteratively thresholded, and final mask gradient image is determined;According to the target image, by Gaussian difference and fixed threshold segmentation, Gaussian segmentation image is determined;The Gaussian segmentation image and the final mask gradient image are merged, and defect region is determined;The defect feature of the defect region is extracted;According to the defect feature, by deep residual convolution network, defect classification is carried out, and the defect category of the PHEMT is output.The application can more accurately determine defect region, avoid the problem that defect region is not complete, improve the accuracy of defect region identification, ensure the integrity and accuracy of defect region.

The application discloses a high-voltage cable grounding loop resistance prediction method based on data driving, collects historical grounding loop resistance detection records of high-voltage cable sections with different static attribute labels under the same voltage level; adopts iterative threshold variational mode decomposition (ITVMD) to decouple irregular interval multi-loop resistance time series into a plurality of intrinsic mode components with different frequency characteristics; splices frequency domain characteristics of loop resistance mode components with static attributes to construct a comprehensive feature vector, and clusters the comprehensive feature vector through a K-means algorithm; combines a two-way delay embedding transformation (TDT) algorithm, and based on Tucker decomposition, adopts a block Hankel tensor autoregressive integrated moving average (BHT-ARIMA) multi-sequence prediction model to predict the high-voltage cable grounding loop resistance. Compared with the prior art, the application effectively solves the technical problems of low prediction accuracy caused by sparse grounding loop resistance data, complex coupling and missing values, and realizes accurate prediction of the loop resistance.

This invention discloses an anomaly cleaning method for multibeam bathymetry point clouds, belonging to the field of data cleaning technology. It includes: acquiring multibeam bathymetry data to obtain a multibeam bathymetry point cloud dataset; performing data augmentation on the multibeam bathymetry point cloud dataset based on a depth-beam adaptive augmentation method to obtain a training dataset that conforms to the physical error laws of multibeam bathymetry; constructing a planar neighborhood local self-attention network to detect point cloud anomalies; training the planar neighborhood local self-attention network using the training dataset; and using the trained planar neighborhood local self-attention network, employing a hierarchical iterative threshold strategy to clean the original point cloud to be detected. The samples generated by this invention more closely approximate the real marine environment, thus enabling the trained model to possess strong generalization capabilities across water depths and incident angles. Simultaneously, modeling the Ping-Beam structural features of the point cloud effectively achieves automated, high-precision processing of multibeam data under complex sea conditions.

The invention relates to a data cleaning method combining space-time clustering and an iterative threshold shrinkage algorithm, belongs to the field of data cleaning and preprocessing, and solves the problems of inaccurate anomaly detection, poor adaptability and low calculation efficiency in existing space-time power data cleaning. Comprising the following steps: S1, acquiring to-be-cleaned original power data in a low-voltage marketing and distribution fusion scene; the original power data comprises time layer data and space layer data; s2, subdividing the time layer data into abrupt change layer data and stationary layer data by using a clustering algorithm, and subdividing the space layer data into high-density layer data and low-density layer data; s3, performing data cleaning on the abrupt change layer data, the stationary layer data, the high-density layer data and the low-density layer data by using an iterative threshold shrinkage algorithm to obtain cleaned layered data; s4, calculating a layered residual error to obtain a final residual error; if the final residual error is smaller than a preset residual error threshold value, clean data are output; otherwise, optimizing the control parameters and returning to the step S2. And efficient and accurate power data cleaning is realized.

The invention discloses an all-sky imager aerosol optical parameter inversion algorithm constructed based on machine learning, and belongs to the field of meteorological observation. According to the method, an iterative threshold segmentation method and threshold self-optimization are used, a clear sky extraction quantitative standard is established for the first time, and the problems of manual dependence and poor consistency are solved; according to the method, an equal zenith angle scanning mode is simulated, multi-dimensional feature extraction is carried out from a sky area picture, aerosol scattering information reflecting multi-angle and multi-position information related to an aerosol scattering phase function and multi-angle information of total brightness is obtained, and key support is provided for SSA inversion; according to the method, an integrated regression model is constructed by taking the multi-dimensional features as input and AOD and SSA as output, a spatial difference and time difference matching sample is set, synchronous output is realized, a sample standard is defined, and efficiency and precision are greatly improved. The method is applied to the fields of atmosphere remote sensing, climate mode research, air quality evaluation and the like.

The application discloses a machine learning-based all-sky imager aerosol optical parameter inversion algorithm and belongs to the meteorological observation field.The application uses an iterative threshold segmentation method, optimizes a threshold automatically, first constructs a clear sky extraction quantitative standard, and solves the problems of artificial dependence and poor consistency.The application simulates an equal zenith angle scanning mode, performs multi-dimensional feature extraction from a sky area picture, obtains scattering information of aerosols which reflects multi-angle, multi-position information related to an aerosol scattering phase function and multi-angle information of total brightness, and provides key support for SSA inversion.The application sets up a regression model with multi-dimensional features as input and AOD and SSA as output, sets spatial difference and time difference matching samples, realizes synchronous output, makes the sample standard clear, and greatly improves efficiency and precision.The application is applied to the fields of atmospheric remote sensing, climate model research and air quality assessment.

This invention discloses a cement velocity inversion method based on arrayed ultrasonic Lamb waves, belonging to the field of oil and gas well exploration and development technology. It solves the technical problems of low computational efficiency and limited applicable velocity range in cement velocity inversion algorithms. The method includes: performing spatiotemporal preprocessing on the array signal to obtain the target guided wave mode, and extracting the complex dispersion curve of the target guided wave mode; constructing a three-layer planar medium model and deriving the dispersion matrix; setting the acoustic parameters of the well fluid and casing, generating multiple sets of initial value sequences for the parameters to be inverted, and calculating the determinant of the dispersion matrix; constructing a loss function, iteratively updating the parameters to be inverted and repeating the process until the number of iterations reaches a preset iteration threshold; traversing all sets of initial value sequences and extracting the iteration parameters corresponding to the round in which the loss function reaches its minimum value. This invention broadens the applicable working conditions and universality, facilitates automated and batch processing, and improves computational efficiency.

The present application relates to a kind of data cleaning method combined with space-time clustering and iterative threshold shrinkage algorithm, belong to data cleaning and preprocessing field, solve the problem of existing space-time power data cleaning inaccuracy, poor adaptability and low computing efficiency of abnormal detection.It includes steps S1, obtaining the original power data to be cleaned in low-voltage camp distribution fusion scenario;Original power data includes time layer and spatial layer data;Step S2, using clustering algorithm, time layer data is subdivided into mutation layer and stable layer data, and spatial layer data is subdivided into high-density layer and low-density layer data;Step S3, using iterative threshold shrinkage algorithm, respectively, mutation layer and stable layer, and high-density layer and low-density layer data are cleaned, to obtain cleaned each layered data;Step S4, calculate the residual error to obtain final residual error;If final residual error is less than preset residual error threshold, then output clean data;Otherwise, optimization control parameter returns step S2.Realize efficient and accurate power data cleaning.

The invention discloses an underwater acoustic communication preamble signal detection method, device and system, and belongs to the field of underwater acoustic communication. The underwater acoustic communication preamble signal detection method comprises the following steps: after a signal is received, carrying out windowing processing on the received signal by adopting a sliding window; performing a hard iteration threshold algorithm on the data in each window to obtain a reconstruction coefficient of each window; performing signal reconstruction based on the dictionary matrix and the reconstruction coefficient, and calculating the energy concentration degree of each window after signal reconstruction; and determining whether a preamble signal exists in the received signal based on the energy concentration degree. According to the scheme, the hard iteration threshold algorithm is adopted to replace a matching tracking algorithm in the prior art, on the basis of ensuring that the detection performance is not changed, the calculation amount and the processing time consumption are greatly reduced, and the requirements for real-time performance and low power consumption of underwater acoustic communication are better met.

The application provides a terminal device and a music generation method. The method can set an evaluation index function and a constraint condition of music generation, and create an initial population, wherein the initial population includes a first population and a second population. A first algorithm is used to iteratively evolve the first population to iteratively search for an optimal individual in the first population, and a second algorithm is used to iteratively evolve the second population to iteratively search for an optimal individual in the second population. An exchange operation is performed on individuals in the first population and individuals in the second population at intervals of a preset iteration number. Finally, target music is generated according to the results of the iterative evolution, wherein the target music is an individual whose fitness function value is greater than a preset fitness threshold when the iteration number is equal to an iteration threshold, and the fitness function value is calculated according to the evaluation index function and the constraint condition. The method generates music in a multi-population cooperative evolution manner, can more comprehensively search the solution space, and thus improves the quality and speed of music generation.

The application belongs to the field of electric power automation, and discloses a kind of artificial intelligence model training method, device and system based on federal learning, comprising: obtaining initial artificial intelligence model and sending to each computing node;Receive the initial artificial intelligence model sent by the center node as local artificial intelligence model;Iterative update step is carried out until the preset update iteration threshold or the current artificial intelligence model meets the preset condition, the current artificial intelligence model is used as the training completed artificial intelligence model, and is sent to each computing node;Iterative gradient calculation step is carried out until the preset gradient calculation iteration threshold, and the training completed artificial intelligence model sent by the center node is received, or, iterative gradient calculation step is carried out until the training completed artificial intelligence model sent by the center node is received, and the training completed artificial intelligence model is used to update local artificial intelligence model. Through the asynchronous model parameter updating method, the computing power utilization rate of high-performance computing node is improved, and the model training speed is improved.

The application discloses a kind of object three-dimensional topography measurement methods based on Unet convolutional neural network adaptive spatial filtering.The computer simulation generates random holographic interferogram conversion into frequency spectrum chart;Iterative threshold segmentation processing obtains frequency spectrum binary segmentation chart;Unet convolutional neural network is built;Frequency spectrum chart is input and frequency spectrum binary segmentation chart is label and is trained;The holographic interferogram of the object to be measured is converted into test frequency spectrum chart, and the object image spectrum center coordinates are determined by phase information;Test frequency spectrum chart is input into the network trained and output frequency spectrum binary segmentation chart, and the object image spectrum region binary segmentation mask is extracted according to the object image spectrum center coordinates in test frequency spectrum chart, and the object image spectrum filtering chart is obtained by filtering test frequency spectrum chart, and the three-dimensional topography chart of the object to be measured is reconstructed.The method has the excellent performance of threshold iterative segmentation method and the rapid processing capacity of neural network, which is beneficial to realize the high-quality and rapid reconstruction of object three-dimensional topography.

The application belongs to the technical field of WSN and discloses a WSN target coverage optimization method based on an improved swarm intelligence algorithm. In view of the problems that the optimal scheme is obtained slowly, is prone to local optimum, and coverage redundancy exists, distribution is uneven, and thus resources are wasted and data quality and reliability are affected due to random deployment of the existing WSN node, the social spider algorithm containing an elite exclusion operator is used to solve the above problems. The method is used for constructing a coverage model, calculating the fitness of a spider individual after initialization, adjusting the position according to the position difference and the update parameter of the optimal spider, determining the elite individual and the exclusion degree during iteration, generating a new population through the exclusion operation, and ending the algorithm when the set iteration threshold is reached. The method improves the coverage efficiency and quality, avoids local optimum, converges fast and is stable in optimization, and provides an efficient scheme for WSN target coverage.

A method for oil spill area detection without human interaction is proposed. This scheme is based on image saliency analysis and an adaptive iterative thresholding method for SAR image oil spill area detection. In this scheme, image saliency detection is introduced into SAR oil spill detection, and then the relationship between the adaptive iterative threshold and saliency is utilized to accurately extract the oil spill area.

The invention provides a self-adaptive fast two-dimensional angle super-resolution method for a real aperture phased array radar, and relates to the field of real aperture radar super-resolution imaging. The method comprises the following steps: modeling azimuth and pitching two-dimensional echoes into convolution of a two-dimensional antenna beam along azimuth and pitching sampling sequence and a target reflection function along azimuth and pitching sampling sequence; under a regularization framework, introducing azimuth and pitching two-dimensional sparse constraints of a target, and converting a deconvolution problem into a parameter estimation problem under the regularization framework; and directly solving a two-dimensional inversion problem by adopting an efficient solving method and a two-dimensional rapid iteration threshold shrinkage algorithm. The method solves the problems that a traditional regularization method is low in processing speed and cannot achieve rapid and effective high-resolution imaging. And the real-time performance of the two-dimensional angle super-resolution is greatly improved.

The application relates to the technical field of image processing, in particular to a PCB gold finger plug external size detection method and system, which comprises the following steps: collecting a plug area image, performing iterative threshold segmentation on the plug area image, and obtaining each highlight connected domain of each threshold; according to the gray value and the contour of a single highlight connected domain of a single threshold, obtaining a complete segmentation factor of the single highlight connected domain of the single threshold; combining the area and the spatial position of a plurality of highlight connected domains of the single threshold, obtaining a complete segmentation coefficient of the single highlight connected domain of the single threshold; combining the area change and the spatial position change of the highlight connected domains between adjacent thresholds, obtaining the complete segmentation degree of each highlight connected domain of each threshold, and obtaining the external size of the gold finger plug of the PCB gold finger area detection board according to the complete segmentation degree. The gold finger area in different positions is completely segmented through the iterative threshold segmentation mode, and the accuracy of detecting the gold finger size is improved.

The invention provides an identity verification method based on fingerprint identification. The identity verification method comprises the following steps: acquiring a fingerprint through a fingerprint collector; the collected fingerprints are preprocessed, including a fingerprint image enhancement method, an image correction method, an iterative threshold method for image segmentation, a two-dimensional FFT transform filtering method and the like, so that the fingerprint images can be better matched. And then feature extraction is carried out on the preprocessed image, a new algorithm based on a binary tree structure is provided to rapidly and effectively process a large amount of two-dimensional discrete data, and the quality of the original image can be maintained to a great extent. A directed graph and an undirected graph are utilized to design and realize a rapid automatic fingerprint positioning algorithm, and a matching result can be obtained through calculation. Aiming at poor image preprocessing effect and low fingerprint image matching precision, the invention provides a rapid automatic fingerprint positioning algorithm, and the accuracy of image matching is effectively improved.