157results about How to "Recognition time is short" patented technology

The invention discloses a machine learning based text identification method of an identity card image belongs to the technical fields including image processing, machine vision and neural network, and solves the problems of long identification time, low identification accuracy and low rotation and distortion resistance when OCR in the prior art is used to identify identity card images automatically in a complex background. The method comprises that a shot image is obtained and preprocessed, and the identity card image is separated from a complex background image in the preprocessed image; character area detection is carried out on the detected identity card image, and character cutting is carried out on a detected text area to obtain single characters; and a deep learning based character identification model is used to identify the cut characters, and an identification result is output. The method can be used to identity text in the identity card image.

The embodiment of the invention discloses a user authentication method and a user authentication device. The user authentication method comprises the following steps: receiving first data information of a user collected by a terminal and comparing the first data information with pre-stored data information corresponding to the user, so as to obtain a first similarity after comparison; if the first similarity does not exceed a preset first similarity, receiving second data information of the user collected by the terminal and comparing the second data information with the pre-stored data information corresponding to the user, so as to obtain a second similarity after comparison; if the second similarity exceeds a preset second similarity, determining that the user passes authentication; and if the second similarity does not exceed a preset second similarity, determining that the user authentication fails. According to the user authentication method and the user authentication device disclosed by the embodiment of the invention, two ways of voiceprint recognition and face recognition are adopted for survival authentication, and therefore, the recognition accuracy rate is high, the recognition time is short, and the security is high.

The invention relates to a hyperspectral data classification method based on a multi-layer convolution network and data organization and folding. The method comprises: step one, pretreatment is carried out before expanding and classification of three-dimensional hyperspectral data and a data matrix including effective spectrum information and a tag vector are obtained; step two, feature dimension expanding is carried out on the data matrix, and column-based folding and reorganization are carried out on a feature dimension to obtain a reorganized three-dimensional hyperspectral data input matrix; step three, a multi-layer convolution network structure parameter and an initial value are set; and step four, a feature and an error are calculated layer by layer by using forward propagation and BP algorithms, a network weight and a bias value are updated, iteration is carried out continuously to obtain a network stablity parameter, and then a network model for classification and a parameter for classification are obtained. Compared with other methods, the provided method has advantages of clear principle, clear structure, short identification time, and high detection identification rate; and the method being an effective classification method for hyperspectral data is suitable for rapid target detection and classification identification application of hyperspectral images.

The invention provides a map element extraction method of an automatic mapping robot, and belongs to the technical field of navigation. The method includes steps: obtaining first map element information of a to-be-detected environment, and constructing a point cloud contour map of the to-be-detected environment, wherein the first map element information includes point cloud information of the position of the to-be-detected environment; detecting a visual mark of the to-be-detected environment at a set position, and reading second map element information included by the obtained visual mark; and marking the second map element information in the point cloud contour map, and obtaining a map including specific semantic information, wherein the second map element information comprises the specific semantic information of the map element corresponding to the position of the visual mark. According to the method, point cloud mapping and semantic marking can be automatically completed, a high-precision semantic map can be obtained, the cost is low, the implementation is easy, and the extensibility is good.

The invention discloses a method for recognizing the gait of a thigh amputation subject and relates to a prosthesis which is not transplanted in a human body. The method comprises the following steps of acquiring multichannel stump surface electromyographic signals of the thigh amputation subject at different gaits, and preprocessing the multichannel stump surface electromyographic signals; performing eigenvalue extraction on the acquired multichannel stump surface electromyographic signals to construct corresponding feature vectors; recognizing the gaits of the thigh amputation subject by using an improved clustering algorithm with a supervision Kohonen neural network. The invention relates to a method for recognizing walking modes of exoskeletons based on the electromyographic signals; five typical gaits of a thigh amputation patient, such as flat ground, upstairs, downstairs, upslope and downslope are accurately classified, and the defects in the prior art that the recognizing results of the walking modes are lagged, effective road condition recognizing is in shortage, and coordinate motion between men and a machine cannot be realized are overcome.

The invention discloses a fish fine-grained classification method based on deep learning. The method comprises the steps of preprocessing an acquired image; carrying out feature extraction by using adeep neural network; constructing a feature pyramid network for regional proposal; carrying out cutting and feature extraction on a proposed area, then carrying out primary classification by utilizingextracted features on one hand, inputting the classification accuracy into an area proposal network as a supervision signal on the other hand, fusing the features and full-graph features, sending thefused features into a full-connection layer for classification, and outputting a final classification result. According to the invention, when an existing object classification technology carries outa fine-grained classification task, a fine-grained classification task is carried out; due to the problems of complex environment, fine inter-class difference between classes and low accuracy causedby large intra-class difference, when the method is used for performing fine-grained classification on the fish images under the complex background, the identification time is short, the identification accuracy is high, extra label information is not needed, and the method is suitable for popularization and application.

The invention discloses a Hilbert-Huang Transform (HHT)-based high-frequency combined coding steady state visual evoked potential brain-computer interface method, which comprises the following steps of: connecting hardware, expressing different targets by using nn stimulation sequences formed by time sequence permutation and combination through n high-frequency stimulation frequencies, coding the stimulation sequences to form a code base, then executing electroencephalogram data characteristics by adopting a Hilbert-Huang Transform (HHT)-based variable frequency electroencephalogram signal characteristic extraction method, acquiring a Hilbert time-frequency graph of the electroencephalogram data, and comparing the extracted electroencephalogram data characteristics with the code base by using a local frequency spectrum extreme value target identification method to realize quantified target identification accuracy. The method has the advantages of simple operation, little electrode number, more target number, reduction of testee fatigue, and reduction of probability of inducing testee epilepsy.

The invention discloses an instrument pointer jittering recognition method based on a support vector machine during interment detection. The instrument pointer jittering recognition method based on the support vector machine during interment detection solves the problems that due to the facts that the accuracy degree of an instrument is decreased and the instrument jitters, errors are caused for the industry, the exertion time of a control variable is long, the magnitude is not accurate, and an existing mode is not accurate or not enough in recognition accuracy. The method comprises the steps of (1) distinguishing subclasses of the instrument, (2) obtaining a training sample, (3) preprocessing the training sample, (4) obtaining a feature space, (5) generating a training model, (6) obtaining the optimal SVM training model, (7) recognizing and segmenting an image to be tested, (8) determining whether an instrument pointer jitters or not, and the like. The instrument pointer jittering recognition method is applied to the field of jittering of the instrument pointer.

The invention belongs to the technical field of the palm print identification, and relates to a palm print identification method based on contour and edge texture feature fusion. The method comprises the steps of firstly preprocessing an image, disintegrating a palm print image into an image contour and an image edge texture by utilizing average value filter, carrying out characteristic extraction by adopting a grey level histogram and differential box dimension, and finally carrying out matching by utilizing the universal chi-square distance. The method is simple, the principle is scientific, the identification time is short, the identification precision is high, the palm print can be identified in real time, and a quick and effective identification method is provided for the identity identification.

The invention provides a method and a system for identifying human facial expression. The method for identifying the human facial expression comprises the steps of utilizing ASM (active shape model) algorithm to extract a shape model and feature points of a to-be-identified facial image, then, utilizing Gabor wavelet transform algorithm to extract amplitude texture characteristics of the feature points; obtaining a to-be-identified eigen face through projection for the amplitude texture characteristics; and identifying expression of a to-be-identified human face by comparing similarity between the to-be-identified eigen face and a sample eigen face. According to the method and the system for identifying human facial expression, the ASM algorithm is combined for extracting the feature points, so that identification through the Gabor wavelet transform algorithm can be performed only by comparing textural characteristics of the feature points. Compared with the existing facial identification technology only adopting the Gabor wavelet algorithm, the method and the system for identifying the human facial expression has the advantages of small data processing capacity, short identification time, low internal memory loss and the like.

The invention discloses a battery RUL prediction method based on battery capacity decline track linear transformation. The method comprises the following steps: S1, reading historical capacity data ofthe power battery, linearizing a non-linear capacity data track through Box-Cox transformation, and obtaining a Box-Cox transformation coefficient and the historical capacity data subjected to Box-Cox transformation; S2, fitting the historical capacity data subjected to Box-Cox transformation and the corresponding cycle index by using a primary polynomial model; S3, identifying model parameters of the primary polynomial model through a Parameter Estimation tool in Matlab / Simulink; and S4, inputting a set capacity threshold value, predicting the number of cycles reaching the capacity thresholdvalue through the primary polynomial model after parameter updating, and outputting the number of cycles. According to the method, the step of analyzing the offline aging data of the power battery isomitted, and resources and time are greatly saved.

The embodiment of the invention provides a target recognition method and device based on a three-dimensional point cloud and a storage medium. The method comprises the following steps: generating a three-dimensional pose LUT of a model; generating a three-dimensional point cloud pyramid; clustering a 3D pose; performing target recognition based on point cloud pyramid. By adopting the method of theembodiment of the invention, the identification and grasping of the disordered target in industry can be well solved, and the position and the posture of the target can be quickly recognized in the current three-dimensional point cloud scene by using the target stl file to be recognized, the recognition time is short, and the recognition accuracy is high.

The invention discloses a motion cycle analysis-based method for identifying an abnormal human behavior. The method is characterized by comprising the following steps of: acquiring human motion behavior areas to form a human motion sequence; computing one-dimensional variable curves and curve frequency spectrograms of the human motion behavior areas in the sequence; judging whether the frequency spectrograms meet a periodical condition, and if all the three frequency spectrograms do not meet the periodical condition, determining that the human behavior is the abnormal behavior; if only one frequency spectrogram meets the periodical condition, determining that the human behavior is the approximate periodical behavior; extracting one motion cycle unit of the approximate periodical behavior; performing the R transform characteristics extraction of the motion cycle unit; sending the characteristics to a single-state hidden Markov model classifier for identification; and giving an alarm when the abnormal behavior occurs. The invention also discloses a motion cycle analysis-based device for identifying an abnormal human behavior, and the device is characterized by comprising a central processing mechanism, a motion cycle extracting mechanism, a motion characteristics extracting mechanism, a classifier and an abnormal behavior alarming device. The device can identify the abnormal human behavior in a higher rate of identification and give an alarm.

The invention provides an intelligent guiding system based on image recognition and used for the blind, belonging to the technical field of artificial intelligence. According to the invention, a userterminal comprises an information collection module which collects video image data, voice navigation instructions, real-time motion data and real-time position information in front of a person with visual impairment in real time. A cloud server comprises a cloud navigation module for generating a navigation route according to the voice navigation instructions and the real-time position information, and a cloud decision module used for detecting a target object in the advancing direction of the visually-impaired person according to a first target detection model and generating a cloud feedbackinstruction to guide the visually-impaired person to advance in a voice broadcast mode. The user terminal further comprises a local decision module which detects the target object in the advancing direction of the visually-impaired person according to a second target detection model and generates a local feedback instruction to guide the visually-impaired person to advance in a voice broadcast mode when the cloud decision module does not respond in time. According to the invention, various obstacles and traffic lights can be effectively identified, and visually-impaired people can be assistedin accurately perceiving a surrounding environment.

The invention discloses a motor mechanical parameter identification method and system based on a high-order sliding-mode observer. The method comprises the following steps of operating a motor under two working conditions of the same acceleration and different rotating speeds, obtaining output signals of the high-order sliding-mode observer, and then calculating a friction coefficient estimation value; eliminating the influence of a friction coefficient error, obtaining output signals of the high-order sliding-mode observer when the motor is operated in two working conditions of different acceleration, and then calculating a torque inertia estimation value; and eliminating the influences of the friction coefficient error and a torque inertia error, obtaining an output signal of the high-order sliding-mode observer when the motor is operated in a working condition, and calculating a load torque estimation value. According to the method, the mechanical parameter identification is realized, the accuracy of a mechanical parameter identification result is improved, and the application value of the mechanical parameter identification is improved.

The invention relates to a thunder identification method, including the steps of: performing framing on an originally input audio signal; converting each frame of audio signal to obtain an audio amplitude spectrum, calculating power of a lower frequency band, and adopting a double-threshold endpoint detection method to identify an audio effective segment; calculating power of an audio signal of the effective segment, and judging a relation between a proportion of the power of the lower frequency band a predetermined threshold value, if the proportion is larger than the predetermined threshold value, judging the audio signal to be a thunder signal, and otherwise judging the audio signal to be a noise signal. The thunder identification method provided by the invention can not only overcome interference of environmental noise, but can also be suitable for thunder signals of different forms, and satisfies requirements of a thunder three-dimensional positioning system in accuracy and real-time performance.

The invention relates to the field of medical image processing, in particular to a urine visible component recognition method based on an improved Alexnet model. The method comprises the following steps: 1, acquiring and expanding an image data set, and constructing a urinary sediment image training set and a test set; 2, constructing a urine visible component recognition network model based on anAlexnet network model; 3, setting training parameters of the urine visible component recognition network model; 4, training a urine visible component recognition network model based on an Alexnet network model; 5, testing a urine visible component recognition network model based on the Alexnet network model. The method is improved on the basis of an Alexnet network model and reduces the number ofnetwork training parameters. The method can automatically extract image features, has the characteristics of high recognition rate, short recognition time and strong generalization ability, and has important application prospects for assisting medical diagnosis and reducing the burden of doctors.

The invention discloses a power transformer magnetizing-inrush-current identification method based on an MEEMD algorithm, and belongs to the field of power system relaying protection. Through collecting differential current signals on two sides of a transformer, MEEMD decomposition is firstly carried out to obtain a set of approximately stationary IMF signals; and then a first IMF signal is extracted from a result of decomposition to carry out Hilbert transform to obtain instantaneous frequency thereof, a sudden-change point of the instantaneous frequency can reflect a sudden change of a phaseof the current signal, and accurate identification on magnetizing inrush currents and internal faults of the transformer can be realized according to a time interval between two adjacent sudden-change points of the instantaneous frequency monitored in real time. The method is not impacted by inrush-current dead angles, closing initial-phase-angles and non-periodic components, and can accurately,reliably and quickly distinguish the magnetizing inrush currents and internal short-circuit faults of the transformer.

The invention discloses a transient component based short circuit fault phase selection method comprising the following steps: obtaining the three-phase traveling wave current signals of n detection points preset in case of power line failure wherein n is an integer greater than or equal to one; according to preset determining rules, choosing targeted detection points and extracting the three-phase traveling wave current signals of these targeted detection points; building a phase-model transformation matrix; performing phase-model transformations to the three-phase traveling wave current signals of these targeted detection points for wire model 1, wire model 2 and wire model 3; carrying out wavelet transformation to the wire model 1, wire model 2 and wire model 3 respectively for their limit values; and according to the preset relations between the limit values of the wire model 1, wire model 2 and wire model 3, outputting the result of selection for phase in fault. The method obtains traveling weave fault current when a power transmission line outages. Through corresponding analysis, the phase in fault can be selected so as to identify the characteristics type for short circuits in shorter time but with higher accuracy.

The invention discloses a parameter identification method for loan dynamics of industrial robot independent of body dynamic parameter, which includes building the body of a robot and a parameter model of load dynamics, calculating a motion excitation trajectory identified by the parameter of load dynamics, carrying out the motion identified by the load in accordance with the motion excitation trajectory, collecting motion data during the motion, establishing the body of the robot and the parameter model of load dynamics in accordance with the step S1 and the motion data collected in the step S3, dealing with the parameter of loan dynamics by hr, estimating the parameter of load dynamics. The parameter identification method for load dynamics of industrial robot independent of body dynamic parameter constructs linear models of the torque in the drive of robot's joint and the parameter of load dynamics, while reducing the complexity of identifying and calculating the parameter of load dynamics.

The invention discloses a smart access control system based on city Internet of Things and a security system. The smart access control system comprises a cloud server, an agency server, a motor vehicle access control system, a non-motor vehicle and person access control system, a cell gate access control system, a building anti-climbing system, a three-dimensional real scene map and a GIS (geological information system), wherein the motor vehicle access control system, the non-motor vehicle and person access control system, the cell gate access control system and the building anti-climbing system are respectively connected with the cloud server; the cell gate access control system comprises an IC (integrated circuit) card access control system, a video telephone access control system, a human face recognition access control system, a mobile phone MAC (media access control) address access control system and a palm vein access control system. The smart access control system and the security system have the advantages that the communities in the city Internet of Things are displayed in the three-dimensional real scene map, then the authority management is performed on persons, motor vehicles and non-motor vehicles in the communities, and the entry and exiting at the entrance and cell gates of the communities can be monitored, so as to improve the intelligence management of the community, and facilitate the public security system to supervise and manage.

A shift control ECU includes the steps of storing a position of a shift lever; starting a count-up timer when the shift lever comes to a neutral position; and recognizing a driver's request for the neutral position when a count-up timer value attains TN(1) if the stored shift lever position is a home position and when a count-up timer value attains TN(2) if the stored shift lever position is not the home position. Here, TN(1) is set to be smaller than TN(2).

The invention discloses a method of quickly identifying thermal state characteristics of a machine tool main shaft, which comprises the following steps: arranging n temperature measuring points on the machine tool main shaft, obtaining temperature values of the n thermometry points in each time under sampling interval Delta t, and obtaining temperature sampling matrix with sampling time m; setting delay time to determine thermal eigenvalue b r (r=1, 2, ..., p); choosing a point with highest temperature in temperature testing points as a key point, and calculating an expression of a temperature rise curve of the key point; calculating root-mean-square error Sigma of a temperature measured value and an estimated value; and obtaining root-mean-square errors Sigma with different sampling time m until the root-mean-square error exists a minimum root-mean-square value. Therefore, a temperature rise curve of a key point corresponding to the minimum root-mean-square value is the curve which can accurately reflect the thermal state characteristics of the machine tool main shaft. According to the method, only about 30 min is required to realize the purpose of quickly identifying the thermal state characteristics, acquire the temperature rise curve of the key value, and shorten the identification time; the result is accurate; the operation is simple.

The invention provides a synchronous calibration method for zero deviation and load parameters of a robot end force sensor. The method comprises the following steps of: establishing a force sensor measuring force model and a torque model; calibrating the identification movement of an industrial robot, designing a calibration motion, and making the industry robot is in a variety of specific configurations, and obtaining the transformation matrix of the corresponding world coordinate system to the sensor coordinate system, so that the sensor zero deviation and the load gravity or gravitational torque are decoupled; according to the relationship between the measured values of the force sensor and the load gravity of the industrial robot in various specific configurations, by collecting the measured values of the force sensor during the motion process, obtaining the force sensor zero deviation and the tool load gravity parameter to be identified synchronously. The method utilizes the specific shape of the robot to obtain a transformation matrix for decoupling the force sensor zero deviation and the load gravity parameter, which is applicable not only to the robot ground installation, but also to the wall installation, the hoisting and the oblique installation.

The invention provides an interactive intelligent table lamp based on virtual reality, and relates to the field of intelligent homes. The interactive intelligent table lamp aims at solving the problems that an existing table lamp is simplex in mode and low in intelligent degree. The intelligent table lamp which interacts with a user through the method of virtual model software or voice or manual operation has the functions of automatically looking for books and faces and automatically regulating the light intensity and the light ray focus according to voice instructions and other functions. By means of controller hardware, on the basis of single-chip microcomputer development with an ATMEGA chip serving as the core and through limited resources of the chip, various ports are expanded, various modules are integrated, and the product is kept compact and the cost-performance ratio is high while the intelligent function of the product is achieved. In addition, by means of a virtual reality system of an upper computer and with the help of a virtual three-dimensional model and the robotics forward and inverse kinematic theory, solution is conducted on movement of the virtual model, and a command for controlling the entity table lamp is generated. The interactive intelligent table lamp is suitable for the field of intelligent homes.

The invention provides an industrial robot load kinetic parameter identification method based on specific joint motion. The method comprises the steps that an industrial robot load kinetic parameter model is set up; a gravity identification track and an inertia identification track are designed according to the industrial robot load kinetic parameter model; load identification motion data gravity identification motion and inertia identification motion are collected and processed according to the gravity identification track and the inertia identification track, and load kinetic parameters can be obtained through motion process joint motion parameters and driving torque parameters. Load weight, the mass center position and rotational inertia parameters can be independently obtained without depending on the kinetic parameters of a robot body through the specially designed joint motion, and the method can be applied to high-performance control of the robot based on the model.

The embodiments of the present invention provide a method for identifying an indoor distribution high-rise cell, a device for identifying an indoor distribution high-rise cell, an electronic device and a storage medium. The method includes the following steps that: an indoor distribution measurement report in a measurement time period is obtained, wherein the indoor distribution measurement reportincludes the identifiers of serving cells and the identifiers of neighboring cells; and a serving cell that satisfies a first preset condition is adopted as a suspected indoor distribution high-risecell, wherein the first preset condition refers to a condition that a distance between a first base station corresponding to the serving cell and a second base station corresponding to a neighboring cell is greater than a first threshold value. According to the method, the indoor distribution measurement report is obtained; the distance between the first base station corresponding to the serving cell and the second base station corresponding to the neighboring cell is obtained; the distance is compared with the first threshold value; and if the distance is greater than the first threshold value, the serving cell is identified as the indoor distribution high-rise cell.

The invention discloses a dynamic signal parameter identification method based on an EKF and an FSA. The method comprises the steps: firstly obtaining a state space expression, containing estimation parameters, in a state vector component; secondly giving a state estimation value and an initial value of a state estimation error covariance, carrying out one-step estimation through employing the EKF in the largest range of iteration, and obtaining an identification result at the next moment; thirdly judging whether the identification result at this moment meets a constraint condition or not; directly carrying out iteration identification again through employing the EKF if the identification result at this moment meets the constraint condition; carrying out optimizing at this moment through employing the FSA if the identification result at this moment does not meet the constraint condition, obtaining the identification result, meeting the constraint condition, at this moment, and carrying out the iteration identification at the next moment on the basis. Through the combination of an EKF (extended Kalman filtering) state estimation method and a fish school optimizing algorithm, the method solves a problem of dynamic signal parameter identification under the actual constraint condition, and enlarging the application range of a fish school optimizing algorithm.

The invention relates to a primary user emulation attack detection method based on channel multipath delay differences. In a present method, when a PUE user can obtain prior information of a PU and has a reconfiguration capability, the detection efficiency greatly decreases. The method provided by the invention is characterized in that channel estimation is carried out to obtain small-scale fading characteristics of channels, and a binary hypothesis detection method is utilized to realize primary user emulation attack detection according to the characteristics. Specifically, the method comprises the steps that: according to a multipath fading model of the channels under a fixed scene, a cognitive radio user utilizes lead code information of the master user to carry out relative operation on received signals and synchronous sequences so as to obtain the multipath delay differences of the channels, then the multipath delay difference of two paths largest in amplitude is selected as a binary hypothesis detection object, and a decision is made. According to the invention, only the lead code information of the primary user is needed, background noise power of the channels are not needed, and normal work is available when the primary user emulation user successfully emulates large-scale and middle-scale fading characteristics of the primary user.

The invention provides a fingerprint recognition switch activated by pressing. The fingerprint recognition module comprises activator, to produce acute position signal and obtain power after receiving activate command that is finger takes action activator can receive; press switch, to run recognition program that is run by the program and produce recognition result if being pressed after obtain power; fingerprint sensing plate, that is on top of press switch and used to obtain fingerprint; and fingerprint recognition processor. When finger forces on fingerprint sensing plate, it presses indirectly the press witch; and as finger is very close to fingerprint sensing plate, ob needed fingerprint for recognition program, then run recognition program.