Patents
Literature

638results about How to "Improve computing efficiency" patented technology

Video fingerprint detecting and video sequence matching method and system based on visual features

The invention provides a video fingerprint detecting method based on visual features. The method comprises the specific steps of segmenting a video sequence according to frame-to-frame correlation, extracting key frames from segments, extracting video features from all key frames, utilizing a pixel point feature dictionary to classify pixel points, respectively blocking all key frames multiple times in different numbers, counting the occurrence times of all elements of the pixel point feature dictionary according to pixel point classifying results in all sub blocks to obtain feature vectors of the sub blocks, splicing the feature vectors of all sub blocks to obtain high-dimensional video fingerprint of the key frames, reducing dimensions of the high-dimensional video fingerprint of the key frame, and connecting the low-dimensional video fingerprints of the key frames of all video segments into a key frame video fingerprint strand according to a time sequence. The invention further provides a video matching method based on the fingerprint detecting method. The video fingerprint detecting method and the video matching method can effectively describe key information of video contents, greatly reduce algorithm complexity under the situation without affecting the matching ratio and effectively improve the detecting efficiency.
Owner:HUAZHONG UNIV OF SCI & TECH

Numerical control interpolation system for real-time generation of curvature-continuous path

ActiveCN102147600AVelocity and acceleration smoothingImprove computing efficiencyNumerical controlLine segmentMachine tool
The invention discloses a numerical control interpolation system for real-time generation of a curvature-continuous path in the technical field of mechanical numerical control processing. The system comprises a man-machine interaction interface module, a decoding module, an interpolation module and a position control module, wherein the man-machine interaction interface module selects a file to be processed and transfers the name of the file to the decoding module; the decoding module extracts the position information of each axis of a machine bed, transmits the position information to the interpolation module and carries out transition on connection points of every two adjacent line segments by a curvature-continuous Bezier curve to generate a curvature-continuous tool path; a maximum deviation value between the generated curvature-continuous path and an original path is smaller than a set error value; speed planning and Bezier interpolation are then carried out; and an interpolation point is output to the position control module at last. The system is high in calculation efficiency of the whole process and simple in implementation of programming and can be applied to a high-speed and high-accuracy numerical control machine bed.
Owner:SHANGHAI JIAO TONG UNIV +1

Facial expression recognition method and facial expression recognition device

InactiveCN107358169AReduce operational complexityImprove computing efficiencyAcquiring/recognising facial featuresFacial expression recognitionEmotion identification
The invention is suitable for the field of information processing, and provides a facial expression recognition method and a facial expression recognition device. The facial expression recognition method comprises the steps that an emotion recognition model based on a convolutional neural network is constructed and trained; a facial image to be recognized is inputted to the emotion recognition model so as to output the emotion class of the facial image, wherein the emotion class includes one of the positive emotion, the negative emotion and the neutral emotion; an expression recognition model corresponding to the emotion class is acquired; and the facial image is inputted to the expression recognition model so as to output the expression class of the facial image. The facial expression is recognized by using the hierarchical mode, and different expression recognition models are selected according to different emotions so that the content required to be memorized of each recognition model can be reduced, the operation complexity of the whole expression recognition process can be reduced and the operation efficiency can be enhanced. Compared with the conventional facial expression recognition method, the facial expression recognition accuracy and recognition efficiency are higher.
Owner:厦门熵基科技有限公司

Dynamic spectrum data processing method based on difference value extraction

The invention discloses a dynamic spectrum data processing method based on difference value extraction and relates to the technical field of spectrum analysis. The dynamic spectrum data processing method comprises the steps of: synchronously acquiring photoelectric volume pulse waves under N wavelengths of a full-wave band of a part to be detected; setting a range of the number of interval points; extracting difference value dynamic spectrum groups corresponding to all numerical values SA within the range of the number of the interval points by adopting a difference value extraction method; and comparing the difference value dynamic spectrum groups corresponding to all numerical values SA and selecting a group with lowest dispersion degree, which serves as the final dynamic spectrum result after being subjected to superposition averaging. The dynamic spectrum data processing method can obtain a large amount of difference value dynamic spectra by virtue of difference value operation, fully utilizes experimental data, enhances computational efficiency, lowers experimental complexity, eliminates the dynamic spectrum with a gross error by virtue of the average effect of the differencevalue dynamic spectra during the elimination process of the gross error, greatly increases the signal to noise ratio of the dynamic spectra and improves the precision of detection on non-invasive blood constituents by the dynamic spectra.
Owner:TIANJIN UNIV

SM2 (streaming multiprocessor 2) digital signature generating algorithm realizing method and device

The invention relates to an SM2 (streaming multiprocessor 2) digital signature generating algorithm realizing method and device. The method includes calculating an SM2 digital signature generating algorithm by coordination of a CPU (central processing unit) and a GPU (graphics processing unit), pre-calculating elliptical curve dots in the SM2 digital signature generating algorithm in the GPU to obtain a result and storing the result in a storage unit, pre-calculating one-added inverse calculation of a user private key in the CPU to obtain a result and storing the result in the storage unit, calling the results obtained by pre-calculation and stored in the storage unit when the CPU calculates a to-be-signed message sequence according to the SM2 digital signature generating algorithm calculating step. Accordingly, digital signature calculation time for to-be-signed messages is shortened, and calculation efficiency of the SM2 digital signature generating algorithm is improved. Further, the CPU can perform one-added inverse calculation on different private keys and store results in the storage unit, and calculation efficiency is further improved.
Owner:DATA ASSURANCE & COMM SECURITY CENT CHINESE ACADEMY OF SCI

Flight path determining method for performing automatic inspection on fan through unmanned aerial vehicle and system thereof

The invention provides a flight path determining method for performing automatic inspection on a fan through an unmanned aerial vehicle and a system thereof. The fan at least comprises a tower, a power generator, a wheel hub and an impeller. The impeller comprises a plurality of blades. The method comprises the steps of establishing a world coordinate system through using a ground center of the tower of the fan as an original O, wherein a Y axis is in a vertical direction, a Y axis is in a vertical upward direction, a Z axis is in a right north direction, and an X axis is a right east direction; performing translation transformation and rotation transformation according to the world coordinate system for generating a power generator coordinate system which corresponds with the power generator, performing translation transformation and rotation transformation according to the power generator coordinate system for generating a wheel hub coordinate system which corresponds with the wheelhub, and performing rotation transformation according to the wheel hub coordinate system for generating a blade coordinate system which corresponds with each blade; and setting a plurality of path points at the front side and/or back side of each blade through the blade coordinate system which corresponds with each blade, wherein each path point comprises a geographical position and camera attitude information, and forming a flight path according to the path points. The method and the system can conveniently realize path point planning and furthermore improve calculation efficiency.
Owner:SHANGHAI CLOBOTICS TECHNOLOGY CO LTD

Point cloud data denoising method based on k near neighborhood division

The invention discloses a point cloud data denoising method based on k near neighborhood division. The point cloud data denoising method based on k near neighborhood division includes the steps: utilizing a table cell method to perform space division on point cloud data; acquiring the neighborhood information of the point cloud data, and completing point cloud data k near neighborhood searching; calculating the average distance between two points in the point cloud data; calculating the value of the influence of any one point in the point cloud data on the point of the k near neighborhood; traversing the average value of the point cloud data influence values, and then, on the basis, setting a threshold compared with the influence; comparing the set threshold with the value of the point cloud data influence to judge whether the point is a noise point; and removing the noise point, and obtaining the denoised point cloud data. The point cloud data denoising method based on k near neighborhood division combines the k near neighborhood information which is determined through space division of point cloud data by means of the table cell method, of any one point, with the Gauss influence function which is taken as the influence evaluation function, can effectively remove the noise points in the point cloud model, and can maintain the characteristic information of the original model and improve the calculating efficiency at the same time.
Owner:CHANGAN UNIV

Wind power disturbance online safety precaution method

InactiveCN102915471AComprehensive calculation of stability assessmentImprove computing efficiencyForecastingSCADAElectric field
The invention discloses a wind power disturbance online safety precaution method. The wind power disturbance online safety precaution method comprises the following steps: acquiring real-time running methods of a power grid and a wind power plant; obtaining possible wind power disturbance of a current power grid in combination with historical data of the wind power plant; comprehensively calculating and evaluating safety and stability situations of the current power grid which is suffered from wind power disturbance; and carrying out online safety precaution of the wind power disturbance. Information of SCADA and information of WAMS are comprehensively utilized to carry out load flow calculation, the result is approximate to the actual situations, and the obtained effect is good. By the wind power disturbance online safety precaution method, high frequency historical monitoring data of the wind power plant are analyzed, the problem of difficulty in obtaining wind power disturbance information is solved, and wind power plant anticipated disturbance accords with actual situation of the power grid. A dynamic safety evaluation and precaution technique is combined with the wind power disturbance online safety precaution method, the power grid which is suffered from the wind power disturbance is subjected to comprehensive safety and stability calculation in distributed and parallel ways, and the calculation efficiency is high. And moreover, by the wind power disturbance online safety precaution method, weak areas of the power grid which is suffered from the wind power disturbance are given, and the calculation results are intuitional and easy to understand, so that practical dispatching and operation are easily guided.
Owner:CHINA ELECTRIC POWER RES INST +1

Closed-loop detection method for indoor scene recognition

ActiveCN106897666AResolve scene confusionImprove computing efficiencyCharacter and pattern recognitionFast algorithmClosed loop
The invention relates to a closed-loop detection method for indoor scene recognition. The closed-loop detection method comprises the following steps that: a current scene image is collected, the FAST algorithm is utilized to extract the feature points of the current scene image, and the LSD algorithm is utilized to extract the feature lines of the current scene image; the ORB is adopted to generate the descriptor vectors of the feature points, and the BRLD is adopted to generate the descriptor vectors of the feature lines; the K-means clustering algorithm is adopted to cluster all the descriptor vectors to generate a visual vocabulary, and the BOW vector of the current scene image is generated by using the visual vocabulary; and the similarity of the BOW vector of the current scene image and the BOW vector of a stored historical scene image is calculated, the consistency of the current scene image is detected, so that whether a closed loop occurs on the current scene image is judged. According to the closed-loop detection method of the invention, the feature lines are additionally adopted in closed-loop detection, sites which have been visited are identified in an indoor environment where feature points are scarce; and since most of feature lines in the indoor environment scene are static, and after the feature lines are introduced, influence on the closed-loop detection algorithm by the change of dynamic objects in the scene is small, and the problem of scene confusion can be better solved.
Owner:SHANGHAI JIAO TONG UNIV

Dispatching method applied to multi-enterable complex manufacturing system

InactiveCN101604409AImprove search efficiency and solution qualityImprove computing efficiencyGenetic modelsAdaptive controlGenetic EvolutionFitness function
The invention discloses a dispatching method applied to a multi-enterable complex manufacturing system, which comprises the following steps: aiming at a typical representative, namely a semiconductor production line as a research object, according to the characteristic of semiconductor production line dispatch, by using regular dispatch as a main line, coding workpiece feeding strategy, workpiece selecting device rule, batch processing dispatch rule and single processing device dispatch rule according to a certain mode; based on considering multi-target optimization on punctual delivery rate, yield and average processing period of workpieces, constructing a fitness function; using the immunogenetic thought to extract a gene segment with large contribution to genetic evolution as a vaccine; and increasing evolution pressure by injecting the vaccine to an individual in a heredity process so as to depend on an immune method to carry out global search to acquire the optimum solution of the dispatch so as to guide the practical semiconductor production and manufacture. The experimental result shows that the immune method has the advantages of simple concept, convenient realization, high convergence speed and the like, and is an efficient re-enterable complex production system optimizing and dispatching proposal.
Owner:BEIJING UNIV OF CHEM TECH

Method and system for estimating position and intensity of ignition source of fire

The invention provides a method for estimating the position and the intensity of an ignition source of a fire. The method comprises the following steps of: dividing a region in which a building structure is positioned into a plurality of blocks; recording static structure information and dynamic structure information of each block, detecting an environment parameter to acquire fire monitoring information and sending to calculation and storage devices; sharing the fire monitoring information, the static structure information and the dynamic structure information of the block by the calculation and storage devices, simulating and forecasting fire simulation information under the intensity of different ignition sources according to the static structure information and the dynamic structure information, and performing partial estimation on the position and the intensity of the ignition source between paired blocks consisting of the block and a neighbor block; and cooperatively processing a plurality of partial estimation values by the calculation and storage devices in the blocks to acquire an overhaul estimation value of the position and the intensity of the ignition source. The invention also provides a device for estimating the position and the intensity of the ignition source of the fire. By adopting a distributed control structure, the calculation efficiency, the robustness and the expandability of the system for estimating the position and the intensity of the ignition source are improved.
Owner:TSINGHUA UNIV +1

Method and device for analyzing nonlinear process of earthquake response of urban building groups

The invention discloses a method and device for analyzing the nonlinear process of earthquake response of urban building groups. The method comprises the following steps of: collecting building data;obtaining models corresponding to the building data according to the building data; establishing a multi-particle shear series model or a multi-particle parallel shear bending coordination model corresponding to the building data according to the models corresponding to the building data; through the multi-particle shear series model or the multi-particle parallel shear bending coordination model,carrying out nonlinear process calculation according to acceleration time history data of earthquake ground motion, which are input by the buildings to obtain nonlinear process calculating results; and obtaining the earthquake damage state and analysis results of each floor of each building according to the nonlinear process calculation results. The method has the advantages that the earthquake damage characteristics of buildings at different heights can be accurately reflected and are closer to the actual earthquake damage; and the method is high in calculation efficiency and simple in modeling method and can be used for accurately and timely predicting and analyzing the earthquake damage of urban typical earthquake scenarios.
Owner:TSINGHUA UNIV

Two-stage serial connection type ISD suspension frame parameter optimizing method

ActiveCN103593506AImprove suspension performanceImprove computing efficiencyBiological modelsResilient suspensionsParticle swarm optimizationFive freedoms
The invention provides a two-stage serial connection type ISD suspension frame parameter optimizing method which includes: building a non-linear two-stage serial connection type ISD suspension frame half-vehicle five freedom degree model containing an inerter; analyzing suspension frame parameter sensitivity according to the degree of influence of each parameter variation on the seat vertical weight acceleration root-mean-square value; using the suspension frame parameters with high sensitivities as the system optimizing variables, and aiming to lower the seat vertical weight acceleration root-mean-square value while vehicle driving safety is guaranteed and suspension frame strike limit probability is controlled; using a particle swarm optimization algorithm to optimize the multi-target optimization model so as to obtain the optimal value of each optimizing variable. The method has the advantages that in order to achieve two-stage serial connection type ISD suspension frame multi-target multi-parameter optimization, system optimizing variables are obtained through suspension frame parameter sensitivity analyzing, then the particle swarm optimization algorithm is used to optimize suspension frame parameters, suspension frame performance is improved, calculation efficiency is optimized, and optimization calculation time is lowered.
Owner:JIANGSU UNIV

Cross-distance-unit and cross-Doppler-unit quick coherent accumulation detection method for radar maneuvering target

ActiveCN109541568AAccumulated gain is highImprove computing efficiencyWave based measurement systemsFourier transformSlow time
The invention relates to a cross-distance-unit and cross-Doppler-unit quick coherent accumulation detection method for a radar maneuvering target, and belongs to the technical field of radar signal processing and detection. The method comprises the following steps that: firstly, carrying out a time reversal operation on slow time dimension data after radar pulse pressure, eliminating a phase odd-order item, and utilizing second-order matching Fourier transform to estimate a second-order phase parameter; secondly, carrying out quick time dimension Fourier transform and second-order phase compensation; thirdly, carrying out a slow time dimension phase difference order reduction operation, and carrying out second-order Keystone transform; and finally, carrying out distance frequency dimensioninverse Fourier transform and slow time Fourier transform to finish cross-unit coherent accumulation, and constructing detection statistical magnitude to carry out maneuvering target detection. By use of the method, the cross-distance-unit and cross-Doppler-unit energy coherent accumulation of the maneuvering target can be realized, and an accumulation gain is high. Meanwhile, the method is a nonparametric search method which is high in operation efficiency and is suitable for engineering application. Through a peak value, the motion parameters of the target can be estimated, and the detection and estimation ability of the radar maneuvering target can be improved.
Owner:NAVAL AERONAUTICAL UNIV +1

Reliability determination method for power distribution network with distributed wind power based on analytical method

The invention provides a reliability determination method for a power distribution network with distributed wind power based on an analytical method. The method comprises the following steps: step I, building a distributed power reliability model; step II, building a wind power output model; step III, building a state transition probability model of a wind turbine generator; step IV, determining the time of every operating state in an output model of the wind turbine generator to obtain the transition probability between the states of the wind turbine generator; step V, determining the fault rate of the wind turbine generator; and step VI, modeling a wind power farm into a conventional unit in a plurality of derated states and performing reliability calculation. The method is suitable for analyzing the reliability of a feeder line radiation type power distribution system and a power distribution system with distributed wind power supply, and the method not only solves the modeling problem of a power supply point in the reliability calculation after a distributed wind turbine generator is connected to the power distribution network, but also solves the problem that the analytical method cannot be used for calculation during reliability analysis of the power distribution network with distributed power.
Owner:STATE GRID CORP OF CHINA +2

Method for recognizing human face based on electrical power system network safety

InactiveCN101276408AImproved face detection and face recognition performanceImprove computing efficiencyCharacter and pattern recognitionNetwork methodLocal feature analysis
A human face identifying method based on network security access of power system, belonging to human face detecting and security identifying field, by using a human face identifying algorithm based on local characteristics analysis and optimization matching, the method according to the invention comprises: estimating several key characteristic points in human face identification via neural network method; using six BP neural networks to carry out characteristic point positioning process for characteristic point position; during the dynamic human face detection, the image function containing human face rapidly varies by the time t, detecting over 20 frames per second; the BP neural network comprises three layers, namely an input layer, a hiding layer and an output layer with same structure, the topology structure of each layer uses integral interconnection mode. With realization of the method, the problem of light compensation in human face detection and identification can be effectively solved, in order to enhance the performance of human face detection and human identification in different light condition. The method also has advantages of high operation efficiency, excellent operating convenience, suitable for wide application.
Owner:长春供电公司

Superpixel-based workpiece surface defect segmentation method

A superpixel-based workpiece surface defect segmentation method has the following key points in a technical scheme: performing image block division on a processed workpiece surface image to obtain m*msub-image blocks of a pixel size; calculating an energy value and an entropy value of a sub-image block gray scale co-occurrence matrix, selecting one image as a standard sample image, comparing energy values and entropy values of remaining sub-image blocks with those of the standard sample image, and obtaining defective images with through screening; utilizing a superpixel SLIC algorithm to segment the defective images into superpixels; adopting a spectral clustering NJW algorithm to cluster the superpixels to obtain a defect segmentation image; and calculating the center of gravity and theregion area of the defect, and according to the positions of sub-image blocks and the focal distance of a camera, calculating to obtain the actual specific position of the defect on the workpiece surface. The method provided the invention calculates from the superpixel level, reduces the large calculation amount based on pixels for a high-resolution large-sized image, has higher calculation efficiency, and at the same time, the interference of background texture and noise in the image are also suppressed.
Owner:XIANGTAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products