55results about How to "Save computing time" patented technology

The invention relates to a visual detecting method based on decoupling detection for circular seams. The method is used for detecting, recognizing and welding the welding seams of circular welding pieces with V-shaped grooves. The method is characterized by comprising the following steps: first, processing pictures and extracting features; and second, decoupling image information. For the method, according to the three-dimensional welding seam feature information extracted by decoupling single frame images, the deviation value of a welding gun in a two-dimensional plane relative to the welding seam center can be calculated, the accurate positioning of welding spots in three-dimensional space can be realized, the decoupling algorithm is performed by monocular vision when the two-dimensional welding seam information is obtained, the tedious matrix transformation when the welding seam information is demarcated is avoided, the demarcating and calculating time is greatly saved, and the real-time performance of a system is obviously improved. On one hand, building a complex imaging model of a camera is avoided, and on the other hand, the demarcating precision, the robustness and the applicability of the camera are improved.

The unmanned vehicle acquires three-dimensional mass point cloud data (point cloud field) through multiline laser radar for hundreds of MB per second, and the data storage space and processing timeliness have extremely high requirement for the computing resource. The invention provides a multiline laser radar three-dimensional point cloud data rapid and effective extraction method without influencing vehicle and lane line feature recognition. The r layer point cloud data of the vehicle multilayer point cloud data in the area of interest of the unmanned vehicle acquired through multiline laser radar are extracted by adaptive distance. Besides, the invention also provides a lane line extraction method through the return light intensity based on distance and angle correction. The requirement of mass point cloud data processing for the computer hardware can be reduced, the storage space and cost can be saved, the timeliness of point cloud data processing can be accelerated and rapid and effective extraction and feature recognition of the vehicle and lane line point cloud data in the area of interest of the unmanned vehicle can be realized. The method is suitable for multiple urban roads so as to be high in anti-interference capability and great in algorithm robustness.

The invention relates to a BIM technology-based three-dimensional bridge template assembling measurement method, and belongs to the technical field of engineering measurement. The method comprises thefollowing steps of: establishing a BIM model; exporting three-dimensional lofting data; converting a plane coordinate; carrying out site lifting by a measurement robot; carrying out data acquisition;and importing the data into the model to carry out recheck. The method has the benefits of saving the measurement data calculation time and error rate, improving the three-dimensional special-shapedbridge template assembling measurement precision, enabling site piece and dislocation to be lower than standard requirements, saving a lot of construction cost, simplifying the measurement work content and improving the template assembling efficiency.

The invention relates to a method for predicating composite material Pi-shaped non-planar glue joint strength based on triangular envelopes. The method includes the following steps that (1), according to parameters of a composite material Pi-shaped non-planar glue joint structure, a geometrical model is established; (2), according to actual working conditions of the composite material structure, loads and boundary conditions of the Pi joint geometrical model are determined; (3), grid partition is performed on the joint geometrical model, and a Pi joint three-dimensional finite element model is obtained; (4), on the basis of the Pi joint three-dimensional finite element model, a finite element stress analysis is performed; (5), according to the linear finite element stress analysis result, a curved edge triangular envelope route of a Pi negative moment steel padding region is set on the basis of the Pi joint three-dimensional finite element model, all stress component values on the curved edge triangular envelope route are extracted, and the average value of the stress component values is calculated and substituted to the failure criterion for predicating the strength. The method for predicating the composite material Pi-shaped non-planar glue joint strength based on the triangular envelopes is suitable for engineering application and can obviously shorten a development cycle and reduce experimentation cost.

The invention discloses a method for identifying hardware gestures, which comprises the following steps of: acquiring touch information of a finger on a touch screen of a mobile terminal, and preprocessing data in the touch information; and identifying the hardware gestures according to the preprocessed data. The invention also provides the mobile terminal for identifying the hardware gestures. According to the technical scheme of the invention, the hardware gestures of the mobile terminal can be quickly and efficiently identified.

The invention relates to a method for detecting a phishing webpage based on nearest neighbour and similarity measurement, which comprises the following steps: a picture of a whole image of a webpage is taken as a start point, and the characteristic of unchanged dimension conversion is extracted; similar characteristics at phishing webpage detection stage are quickly queried, and are then submitted to a machine leaning and matching module to carry out identification; the machine leaning and matching module extracts characteristic data transmitted during a system training stage to carry out training, so that a parameter of webpage similarity threshold can be optimized; during the phishing webpage detection stage, the characteristic data transmitted by the characteristic extracting module is received, the similarity between webpages is calculated, and finally, the phishing webpage is judged according to the webpage similarity threshold; in addition, a sorting method-Bayesian addable regression tree is added to predict suspicious webpages; and the characteristics during the phishing webpage detection process are extracted to be used as an evidence of the phishing webpage detection, so that the high accuracy can be ensured, and simultaneously, the webpage detection time can be remarkably reduced.

The invention provides a method and device for obtaining a unique visitor (UV), relating to the field of a database. The method comprises the following steps of: traversing a primary user log and taking a user id as a standard to stamp an overall label on a dimension of each strip of data under each dimension; by taking the label, a dimension combination and the user id as keywords, collecting and gathering the data to obtain user granularity data; traversing the obtained user granularity data; and taking the label and the dimension combination as the keywords to collect and gather the user granularity data to obtain the UV of each sub-dimension under each dimension. According to the method and the device disclosed by the invention, only an original table is subjected to primary traversing in an HIVE to finally obtain each usual UV under any dimension, so that the calculation resource and the calculation time are greatly saved.

When image junk mails are detected and determined through junk image filtering technology research based on semi-supervision, text and images characteristics are extracted to conduct characteristic processing. Detecting and sorting are conducted by using obtained sorting models, new label sample are added continuously, a sorter is trained, sorting precision is improved, and simultaneously misjudgment is greatly reduced. A large amount of experimental data inspection shows that the method builds a high efficient junk mail webpage filtering system, high accuracy rate is guaranteed and simultaneously processing efficiency is greatly improved, and webpage detecting time is greatly shortened.

The present invention discloses a Kalman filter channel-based pulse stacking correction method which mainly solves the problems in the prior art that the calculation is complex, an energy spectrum distorts, etc. The method comprises the following steps of utilizing a pulse shaper to generate a nuclear pulse signal Sn, obtaining the nuclear pulse signal Sn and then initializing the parameters of the nuclear pulse signal Sn, namely, selecting an initial point where the amplitude X0 of the outputted nuclear pulse signal Sn is 0; carrying out the weak signal removing processing on the nuclear pulse signal Sn, and obtaining the stacked nuclear pulse signals S1n; determining and extracting the stacked nuclear pulse signals S1n; constructing a Kalman filter state and measurement equation, and using a Kalman filter to process the extracted nuclear pulse signals S1n to obtain the nuclear pulse signals S2n; attenuating the nuclear pulse signals S2n, and determining whether to satisfy a removingstandard. By the above scheme, the Kalman filter channel-based pulse stacking correction method of the present invention achieves the purpose of separating the stacked pulse events, and has the very high practical value and popularization value in the nuclear radiation detection, the nuclear electronics technology and the digital signal processing fields.

The invention relates to a composite material Pi-shaped gluing connection structure tensile strength prediction method based on an average invalidation index. The method includes the following steps: (1) a Pi connector geometrical model is built according to geometrical parameters of a composite material Pi-shaped gluing connection structure; (2) determining tensile load and boundary condition of the composite material Pi-shaped connector geometrical model according to the stress condition of the integrated composite material structure; (3) acquiring an accurate Pi connector three-dimensional finite element model through grid encryption based on the Pi connector geometrical model, meanwhile ensuring that grids on key connection faces L, U and B are even and calculating stress distribution of the three-dimensional finite element model under tensile load; (4) extracting positive axis stress component value of nodes on the Pi connector key connection faces L, U and B and calculating the invalidation index Rij on the key connection face; (5) respectively calculating the average invalidation indexes based on the key connection faces L, U and B of the Pi connector; (6) conducting calculation according to the tensile load P0 and the maximum value of the average invalidation indexes to obtain the invalidation strength value P of the obtained connector. The method is applicable to engineering application, can remarkably shorten the developing period of the Pi connector and reduces test cost.

The invention discloses a metal plate surface defect detection method based on machine vision, and relates to the technical field of digital image processing. The method includes: projecting red stripe light to the surface of a plate to highlight defects like pits, bulges and scratches, and acquiring a stripe projection image by using a CCD camera; decomposing a color image, highlighting the colorinformation of the light source and highlighting the color information of the light source; and extracting a stripe center, judging the distortion condition of the line through an algorithm, and reflecting the size of the defect. A stripe lighting mode is adopted, the influence of the mirror reflection effect is eliminated, meanwhile, the display force on the surface defects of the plate can be enhanced, and a defect image with high quality is obtained. For plate defects of different types, different forms and different positions, the detection effect is 100% adaptive, which indicates that the method is applicable to metal plate surface defect detection.

Provided are a method and a device for producing a panoramic deep image. The method for producing the panoramic deep image comprises the steps: (1) shooting according to a main focus distance and a minor focus distance respectively to produce a main focus image and a minor focus image respectively, (2) diminishing the main focus image and the minor focus image respectively, (3) conducting geometric correction on the diminished main focus image and the diminished minor focus image to produce a corrected minor focus image, (4) dividing the main focus image and the corrected minor focus image into a plurality of blocks, (5) judging whether each block in the main focus image needs to be compensated by means of the corrected minor focus image, (6) producing a compensation block by means of an overlap block compensation method and a weight type differential consolidation method, and (7) compensating the main focus image by means of the compensation block to produce the panoramic deep image.

The invention relates to a method and a device for hierarchical clustering, and belongs to the field of data mining. The method comprises the steps of obtaining a data object set to be clustered, wherein the data object set comprises a plurality of classes, and each class corresponds to at least one data object; clustering the data objects corresponding to the first class to obtain a clustering result, wherein the number of the data objects corresponding to the first class is greater than a first preset threshold value, the clustering result comprises a plurality of clusters, and each cluster comprises at least one data object; screening the data objects corresponding to the first class according to the clustering result to obtain representative data objects of the first class; calculating a between-class distance according to the representative data objects of the first class and data objects corresponding to the second class; performing hierarchical clustering on the data object set based on the between-class distance. According to the method and the device for hierarchical clustering, by clustering of the data objects corresponding to the first class, the calculation intensity is reduced, and the calculation time and resources are saved; furthermore, the clustering result is more reliable, and subsequent data analysis is facilitated.

The invention discloses a human motion identification method based on second generation Bandelet statistical characteristics, which mainly solves the problems of complicated characteristic extraction and poor representational capacity in the prior art. The human motion identification method comprises the steps of: 1, converting video in a Weizmann database into a sequence image, constructing a training sample set X and a test sample set T according to a proportion of 8:1; 2, carrying out second generation Bandelet transformation on a single sequence image in the sample sets, sequentially extracting an energy characteristic Ve, an entropy characteristic Vs, a maximum value characteristic Vmax, a minimum value characteristic Vmin, a contrast ratio characteristic Vc, a mean value characteristic Vmu and a variance characteristic Vv of the image, and cascading all the characteristics to be used as final characteristics of the single image; and 3, repeating the step 2 for respectively extracting characteristics of all sequence images in the training sample set X and the test sample set T to obtain a training sample characteristic set X* and a test sample characteristic set T*, and carrying out learning training on the training sample characteristic set X* and the test sample characteristic set T* by using an Adaboost algorithm to obtain a classifying result. The human motion identification method can be used for accurately identifying a human motion, and can be used in video monitoring and video processing of target identification.

The invention discloses a metal target transient electromagnetic scattering analysis method based on delay laguerre polynomials. A target is subdivided through curved surface triangle surface elements. Spatial dispersion and time dispersion are carried out on a time domain mixed field integral equation through a CRWG primary function and the delay laguerre polynomials. A test is carried out through a galerkin method. An order-stepped matrix equation is finally formed, and the transient electromagnetic scattering characteristic of the target can be obtained through resolution of the equation and post-processing. As a delay item related to the spatial position of the target is introduced into a time primary function, the time primary function contains the spatial phase information of the target is adopted, and the target can be subdivided with less curved surface triangle surface elements. The unknown calculation amount is greatly reduced, consumption of an inner storage is effectively reduced, the calculation time is shortened, and important reference documents are provided for accurate analysis of the transient electromagnetic scattering characteristic of an electric large-size metal target.

The invention relates to a method for detecting image-based spam by utilizing the image local invariant feature, which comprises the steps of extracting invariant region feature of spam information in an image by utilizing the accelerated extraction algorithm with the robust feature, further generating a feature vector of the image, estimating parameters of a Gaussian mixture model by using the maximum likelihood algorithm and training a classifier of the Gaussian mixture model. Experiments show that the method can improve the recall rate of the spam and save the program computation time and the space. The classifier based on the Gaussian mixture model is obtained. The realizing method for detecting the image-based spam comprises three modules of the extraction of the image feature, the estimation of the parameters of the Gaussian mixture model and the detection of the image-based spam.

Interpolations for a picture is done after edge detections in a de-interlace processor. By doing so, an amount of computing processes and a total time for the computing are saved; an efficiency of the computing is improved; and, a visual quality of the picture is enhanced.

The invention provides an online combat intention recognition method and device based on incomplete information, and the method comprises the steps: obtaining information data through various types of detection and sensing equipment, and obtaining historical time-varying situation information formed by a continuous tracking signal of each target unit in a time period delta T; performing coding completion compression processing on the historical time-varying situation information to obtain effective input data; inputting into a deep learning model for training to obtain a trained deep learning model; and inputting the current intelligence data into the trained deep learning model to obtain a target intention recognition result. A learner is used for mining a global structure, learning representation of potential shared information, mining more global structures from limited battlefield information and discarding low-level information and more local noise, time characteristics of target intelligence information are considered, and a variable-length time sequence processing model is designed for intention classification learning. And an online intention identification effect under incomplete information is realized.

The invention discloses an improved model predicted control method. A voltage and a 90-degree delayed signal thereof are used in an alphabeta static coordinate system, and a rectifier leg voltage reference value is obtained via a mathematical model of the three-phase three-level eight-switch NPC rectifier, nine spatial voltage vectors of the three-phase three-level eight-switch NPC rectifier are reduced to seven voltage vectors by removing voltage vectors causing a high common-mode voltage; according to an objective function, an optimal spatial voltage vector enabling the optimal objective function is selected; and the obtained optimal spatial voltage vector serves as a switch function of a next control period, the state of a switch tube is controlled, the operation state of the circuit iscontrolled, and a control target of three-phase there-level eight switches in the wind power generation system under the AC input voltage unbalanced condition is achieved. The algorithm is simple, visual and clear, control is simple and accurate, and the practical value is high.

The invention discloses a method for predicting material thermal conductivity on the basis of finite difference method of three-dimensional image, and belongs to the method for predicting the material thermal conductivity on the basis of finite difference method. The method has the basic principle: 1) an image analysis method is used for distinguishing different phases or components in the image; 2) a computer language program is used for reading the position and color information of all pixels into a computer memory; 3) each pixel is constructed into a cell element, and the cell element is endowed with the thermal conductivity according to the component to which the cell element belongs; 4) a stable thermal conductivity equation is dispersed, calculating a heat transfer coefficient matrix is calculated, and a calculation equation set is constructed; 5) a temperature field is obtained by the calculation equation set; and 6) the material equivalent thermal conductivity is calculated. The three-dimensional model is adopted to accurately obtain a thermal conductivity result, and the finite difference method is adopted to shorten calculation time and save memory consumption. The whole process is automatically finished, and the method is very practical for the model with a huge node amount.

The invention relates to the technical field of detection methods, in particular to a roller kiln energy consumption anomaly detection method based on self-adaptive principal component analysis. The roller kiln energy consumption anomaly detection method comprises the following specific steps that (1) firstly, a sample L is collected, the collected sample is standardized, and then a covariance matrix S is calculated; (2) characteristic decomposition is carried out on S, a principal component number k is obtained through calculation by using a CPV method, and a characteristic vector and a characteristic value are intercepted to obtain a load matrix P and a characteristic value matrix lambda; (3) an initial control limit is calculated; (4) a new sample of the sample L is continuously collected after a certain moment, standardizing is carried out, and abnormal judgment is carried out on the new sample; and (5) the steps (1), (2), (3) and (4) are circulated until the number of the samplesneeding to be updated reaches the number of the samples needing to be updated, the samples are updated and abnormal information judgment is carried out. According to the roller kiln energy consumptionanomaly detection method based on the self-adaptive principal component analysis, the abnormal condition of the roller kiln can be effectively detected, and the detection efficiency is greatly improved.

The invention is suitable for the field of computational aero-acoustics and computational fluid mechanics, and provides a physical attribute and data drive coupled streaming acoustic mode decomposition and prediction method.The streaming acoustic mode decomposition method comprises the steps that firstly, a DMD method is adopted for conducting dynamic mode decomposition on an initial velocity field, and a normalized dynamic mode is obtained; carrying out Helmholtz decomposition on the obtained dynamic mode to obtain an acoustic mode speed and a dynamic mode speed; and the acoustic modal speed and the dynamic modal speed at any moment can be predicted based on the obtained acoustic modal speed and the dynamic modal speed, so that the time development process of the acoustic modal and the dynamic modal can be analyzed. According to the method, the advantages of Helmholtz decomposition based on physical attributes and a dynamic mode decomposition method based on data driving are coupled, accurate and efficient decomposition and rapid prediction of the dynamic mode and the acoustic mode of the velocity field in the unsteady flow field are achieved, and the calculation time and cost are saved while the accuracy of the physical attributes is guaranteed.

The invention relates to a semi-supervised graph representation learning method based on fusion of transfer learning and deep learning and a device thereof. The method comprises the following steps: pre-training a graph neural network model through two sub-tasks of a global level and a local level, so that universal representation of input data is learned from unlabeled data; and migrating the pre-trained graph neural network model to the training process of the target task, adding an output layer related to the target task behind the pre-trained graph neural network model, and performing fine tuning on parameters of the pre-trained graph neural network model by using the labeled data to obtain a final graph neural network model. On the basis of saving the manual marking cost, the non-label data and the label data are effectively combined, the generalization ability of the model is improved, the training process of the target task can be simplified, and the purpose of faster convergence is achieved; according to the method, the thought of transfer learning is fully utilized, a large amount of computing resources and computing time can be saved, and the computing efficiency is improved.

The invention provides an industrial part 6D pose estimation-oriented method and a computer readable storage medium. The method comprises the steps of determining a mapping function of a parameter value and a parameter key point and constructing a three-dimensional model library of limited number of size instances; based on the three-dimensional model library, generating a stacked data set throughsimulation of a physical engine and a rendering engine, and generating a parameter key point label and a parameter value label which are used for training a neural network provided by the technical scheme; predicting a spatial offset vector from each point to the mass center and parameter key points of the industrial part to which the point belongs; achieving individual segmentation of the stacking scene through a clustering algorithm, obtaining prediction centroids of individuals, classifying parameter key points of point-by-point prediction, obtianing prediction parameter key points and centroids of the individuals, then calculating point-by-point prediction parameter values, obtaining prediction parameter values of the individuals after classification, wherein the prediction parametervalues are used for calculating parameter key points and centroids of template instances; and obtaining a 6D pose estimation result through a least square fitting method.

An image based adaptive finite element mesh division method belongs to a mesh division method for finite element analysis of a material structure. The method comprises: 1, distinguishing different phases or components in an image with an image analysis method; 2, reading position information of all pixels into a computer memory by utilizing a computer language program; 3, according to the fineness of a spatial structure, reducing an image resolution of a region thick and big in structure and single in component; 4, directly constructing a finite element unit by the pixels, wherein different resolutions cause different mesh densities; and 5, transiting meshes different in density by mesh mapping. According to the method, nodes and units are directly constructed by utilizing images, and solid modeling is skipped, so that system resources can be saved; moreover, a mesh structure can be optimized on the premise of ensuring calculation precision, so that the number of nodes is remarkably reduced, the calculation time is shortened, and the memory consumption is reduced; and the whole process is automatically completed, so that the method is very practical for a model with a huge amount of nodes.

The invention discloses an elliptic curve digital signature method, which comprises the following steps of: calculating an ECDSA digital signature generation algorithm by adopting a division cooperation mode based on a CPU (Central Processing Unit) and GPU (Graphics Processing Unit) heterogeneous structure, deforming a Fermat's small theorem, and optimizing implementation steps of modular inverseoperation of elliptic curve digital signatures in a GPU by using the deformation method; and calculating multiple points of an elliptic curve in the elliptic curve digital signature in batches and calculating results according to GPU; enabling the CPU to calculate messages to be signed according to orders of calculation steps of an elliptic curve digital signature, making full use of the parallelism of the GPU architecture, so that parallelism of a GPU architecture is fully utilized and calculation efficiency of an ECDSA digital signature generation algorithm is improved. therefore, the calculation time of digital signature of the to-be-signed message is saved.

The invention relates to a user default prediction method and device, electronic equipment and a computer readable medium. The method comprises the following steps: obtaining financial data of a user,wherein the financial data comprises a plurality of types of feature data; performing character conversion on the financial data to generate user data; and generating a default category and a defaultprobability of the user through the user data and a user default prediction model, wherein the user default prediction model is obtained by training a distributed gradient boosting decision tree model. According to the user default prediction method and device, the electronic equipment and the computer readable medium, the model training and calculation time can be saved, the grid search time ofa model is saved, and the user default prediction efficiency and accuracy are improved.

The invention provides a method for calculating the estimated stroke width (ESW). The method comprises the steps of obtaining stroke edge information according to a binary image; calculating the stroke width of each stroke edge pixel point in not less than four orientations, with the stroke width of each stroke edge pixel point in the not less than four orientations being the distance from the stroke edge pixel point to another stroke edge pixel point located on a straight line determined by the stroke edge pixel point and the orientations; associating the stroke width of each stroke edge pixel point in the not less than four orientations with each stroke interior pixel point passing the stroke edge pixel point and along the orientations; and for each stroke interior pixel point, selecting the minimum value of the plurality of stroke width values associated with the stroke interior pixel point as the ESW of the stroke interior pixel point. The invention further provides an apparatus corresponding to the method.

In order to solve the problems that trademark image retrieval is large in cardinal number, low in speed and not accurate enough, The invention provides a trademark image retrieval method based on a deep hash method, which comprises the following steps of: extracting a feature vector of a trademark image by using a convolutional neural network, converting the feature vector into a binary hash code,and comparing similarity features of the image by calculating a Hamming distance of the binary hash code. According to the method, the deep hash method is well applied to trademark image retrieval, so that the trademark image retrieval precision is higher, the speed is higher, and the error is lower.