61results about How to "Preserve accuracy" patented technology

A method for transforming three color input signals (R, G, B) corresponding to three gamut defining color primaries to four color output signals (R′, G′, B′, W) corresponding to the gamut defining color primaries and one additional color primary W for driving a display having a white point different from W includes the steps of: normalizing the color input signals (R,G,B) such that a combination of equal amounts in each signal produces a color having XYZ tristimulus values identical to those of the additional color primary to produce normalized color signals (Rn,Gn,Bn); calculating a common signal S that is a function F1 of the three normalized color signals (Rn,Gn,Bn); calculating a function F2 of the common signal S and adding it to each of the three normalized color signals (Rn,Gn,Bn) to provide three color signals (Rn′,Gn′,Bn′); normalizing the three color signals (Rn′,Gn′,Bn′) such that a combination of equal amounts in each signal produces a color having XYZ tristimulus values identical to those of the display white point to produce three of the four color output signals (R′,G′,B′); and calculating a function F3 of the common signal S and assigning it to the fourth color output signal W.

The present invention is a method of hydrodynamics-based gridding (Hydro-Grids) for creating geologic models of subsurface volumes, such as reservoirs. Vertical grid surfaces may be chosen in an unstructured fashion to provide lateral resolution where needed. Lateral grid surfaces are created to represent surfaces of constant geologic time based on simulation of the depositional processes that created the subsurface volume. The values of geologic properties are then specified within each cell created by the intersections of the vertical and lateral surfaces. The geologic data may include, for example, seismic data, outcrop studies, well log data, core data, numerical modeling data, and interpreted stratigraphic surfaces based on seismic data. The modeled geologic properties of the subsurface volume may include, for example, grain size distribution, connectivity, net-to-gross, porosity, permeability and pore pressure.

The invention provides a deep learning image identification method and a deep learning image identification system used for intelligent driving, and a terminal device. The deep learning image identification system comprises a sharing convolutional network, an area segmentation network, and a target identification network. The area segmentation network is used for area classification processing based on a characteristic graph extracted by the sharing convolutional network, and the target identification network is used for target identification positioning processing based on the characteristic graph extracted by the sharing convolutional network. The sharing convolutional network is monitored by using the area segmentation result acquired by the area segmentation network and the target identification result acquired by the target identification network, and the sharing learning of the area segmentation network and the target identification network is completed. An obvious speed advantage on an aspect of multi-task learning is provided, and by comparing with the two independent networks learning individually, the deep learning image identification system has advantages of less consumed time and high efficiency, and in addition, a convolutional layer repetitive operation problem is effectively prevented, and multi-task detection and multi-task identification are completed.

A protection mechanism for force-based touch sensitive input panels or displays comprising a “snap dome”, e.g., a semi-rigid pliable dome positioned over each sensor for increased overload protection. The snap dome is unidirectionally-resilient to a known point of collapse. It therefore imparts a predetermined resistance to compression over a known range of travel along the z-axis, but is unyielding along the x- and y-axis. As the touch panel is depressed toward the standoff, the snap dome resists compression until it collapses, allowing the touch panel to encounter the standoff. The standoff then imparts dead-stop overload force protection to the sensors. This allows for the placement of mechanical stops (such as stop screws) with a less exacting tolerance. Moreover, the snap domes introduce no x- or y-axis force losses / increases, and do not cause tilting of the touch panel. The touch force remains perfectly perpendicular and accuracy is preserved.

A geospatial modeling system may include a geospatial model data storage device containing geospatial model data including a plurality of groups of tree crown vegetation points and a display. The system may further include a processor cooperating with the geospatial model data storage device and the display for displaying a simulated tree trunk beneath a group of tree crown vegetation points, and displaying at least one level of simulated tree branches for the simulated tree trunk. Each tree branch may be positioned based upon determining a branch distal reference point beneath a cluster of tree crown vegetation points from among the group of tree crown vegetation points.

The invention discloses a DOA (depth of anesthesia) monitoring device which comprises an EEG (electroencephalogram) collecting electrode, a pre-arrangement module, a Fourier transform module, an Ac (activity of cortex) calculation module, an As (activity of subcortex) calculation module, a balanced DOA calculation module and a display module, wherein the pre-arrangement module comprises a pre-amplification module and an analog-digital conversion module which are connected sequentially; the Ac calculation module receives output data of the Fourier transform module and calculates subparameters of Ac; the As calculation module receives output data of the Fourier transform module and calculates subparameters of As; the balanced DOA calculation module receives subparameters output by the Ac calculation module and the As calculation module and obtains a balanced DOA with a weighted sum method; and the display module displays the balanced DOA. On the premise that the DOA monitoring accuracy in the state of pure inhalation or intravenous general anesthetics is reserved, the DOA monitoring device has the advantage that the DOA monitoring is more accurate in the state of opioid medicine composite inhalation or intravenous general anesthetics.

The present invention discloses a vehicle abnormal behavior detection and tracking method based on DSRC. The method comprises: a road side personal communication unit is configured to obtain information from a vehicle device, a roadbed computer is configured to calculate the information to obtain the speed of the vehicle, the traffic flow of the road and the traffic density, perform speed abnormity determination through adoption of a threshold value comparison method by a abnormity detection module and determine the vehicle illegal parking and retrograde motion through a design algorithm. Through adoption of a DSRC communication technology between vehicles and roads, the vehicle abnormal behavior detection and tracking method based on DSRC provides a calculation method of microcosmic traffic volume, vehicle speed, macroscopic traffic volume, traffic flow and traffic density so as to solve the problem that the vehicle abnormal behavior is identified and tracked with low cost and accuracy in the car networking environment.

The invention provides an overload proof power battery charging-discharging monitoring control method. The method comprises the following steps: calculating the temperatures of a battery to be testedand the extreme current charging / discharging rates at different SOCs (Systems on Chip) according to a voltage change ratio; determining different temperatures and charging / discharging voltage upper limits at different SOCs according to the maximum current charging / discharging ratio; judging whether failure duration exceeds a preset unexpected duration or not by taking the current charging / discharging ratio under each temperature as an overload current threshold value, and the charging voltage upper limit and discharging voltage lower limit at different temperatures and different SOCs as battery aging failure threshold values, and transmitting an alarm signal; and calibrating the power battery capacity after aging, and calculating a charging voltage upper limit and a discharging voltage upper limit under the current battery capacity. The maximum current loading ratio and voltage working range are taken as monitoring objects of the whole battery system in working, so that the overload proof power battery charging-discharging monitoring control method has the advantages of convenience in operation, high reliability, easiness in judging the working state, recognition of the failure type, and improvement on the system feasibility and reliability.

A geospatial modeling system may include a geospatial model data storage device containing geospatial model data including a plurality of groups of tree crown vegetation points and a display. The system may further include a processor cooperating with the geospatial model data storage device and the display for displaying a simulated tree trunk beneath a group of tree crown vegetation points, and displaying at least one level of simulated tree branches for the simulated tree trunk. Each tree branch may be positioned based upon determining a branch distal reference point beneath a cluster of tree crown vegetation points from among the group of tree crown vegetation points.

The invention relates to a language model training method based on knowledge distillation, a text classification method, a language model training device based on knowledge distillation, a text classification device, electronic equipment and a non-temporary computer readable storage medium. The language model training method based on knowledge distillation comprises a first word vector layer parameter determination step and a language model training step. The text classification method comprises the steps of obtaining a to-be-classified text; based on the to-be-classified text, obtaining a keyword code list of the to-be-classified text through extraction; obtaining a word vector of each keyword corresponding to the to-be-classified text through a language model according to the keyword code list; and through the text classification layer, obtaining a classification result of the to-be-classified text. According to the method, a knowledge distillation method is adopted, the dependence on a labeled sample is reduced while the accuracy of the model is reserved, and the reasoning speed is increased by simplifying the structure of the model, so that the applicability and reliability ofthe text classification method in an intelligent auxiliary secret setting system are improved.

Virtual Reality systems may be used in healthcare and therapy, e.g., focusing on physical and neurorehabilitation. For instance, victims of brain injury may seek treatment to improve, e.g., range of motion, balance, coordination, joint mobility, flexibility, posture, endurance, and strength. VR systems can be used to entertain and instruct patients in their movements while recreating practical exercises to further therapeutic goals. Patient movement data during physical therapy sessions may be valuable to patients and healthcare practitioners. The system may comprise a plurality of body sensors, a VR headset, and a supervisor tablet. The disclosed system may facilitate translation between real world coordinates and virtual world coordinates, assigning sensors to body parts, measuring range of motion of joints and limbs, correcting sensor orientation, recording and presenting therapy data, and generating and animating a 3-D virtual reality avatar in a virtual world to perform activities, among other benefits.

The present invention discloses a multi-scale grain image segmentation based on self-adaptive window fixation and spread. The process comprises the steps as follows: an image block n corresponding to the grain of an image to be segmented is picked up for wavelet transform, and a corresponding HMT model parameter theta n thereof is determined; the corresponding likelihood value of corresponding data block at each wavelet decomposition scale of the image to be segmented and the corresponding likelihood value of the pixel of the image to be segmented are determined respectively and are combined together to find out the likelihood value n<k> required by finial fusion; the likelihood value on fusion widest scale (k is 4) is found out, and the corresponding segmentation result plotting on the scale is determined also; a marking field on the fusion scale k and the physical clustering center of each grain are confirmed; the multi-scale segmentation of the self-adaptive window fixation and spread is used for find out the segmentation result plotting on next fusion scale k minus 4; the finial segmentation result is confirmed by judging whether the fusion scale of the segmentation result plotting is zero or not. The multi-scale grain image segmentation has the advantages of area consistency and good edge positioning performance and can be used for image segmentation comprising grain information.

The invention provides a molecular marker related with the sheep tail fat weight, and a detection method and applications thereof. A sheep ADRA2A gene is subjected to PCR amplification and sequence analysis, people find that one C / A polymorphic site exists on the 217th of an amplified fragment, then a KASPar primer pair is used to detect the polymorphic site of 161 Hu sheep, a least square methodmodel is established, the relationship between genotype and the tail fat weight is analyzed, and the amplified ADRA2A gene fragment can be used as a molecular marker related with the sheep tail fat weight. The provided molecular marker can be used to breed species of low-fat type sheep and low-fat type meat sheep, a genetic engineering method is provided for the improvement of genes related with sheet meat quality, and the method has a high application value.

The invention discloses a method and system for making a WeChat H5 web page. The method mainly includes the steps of establishing a network data connection between a web server and an editing terminal; loading basic production data to the editing terminal; setting the compression ratio of the data according to the network state of the editing terminal; according to user's choices, obtaining production data for editing the WeChat H5 page by a storage unit of the editing terminal and / or a storage unit of the web server; generating the WeChat H5 page according to the acquired production data, user operations and set rules; uploading the above WeChat H5 page to the web server for storing or publishing according to the user operations. In the process of editing, data transmission between the editing terminal and the web server is all compressed so as to reduce the occupation of communication resources by the data transmission, and in the process of compression, data is encrypted to prevent third party tampering, so that the data transmission in the whole process of editing is faster, and the method and system for making the WeChat H5 web page are suitable for use at mobile terminals.

The invention provides a single-phase voltage sag detection method and system based on an EEMD (ensemble empirical mode decomposition) method and a two-point method and application. According to the method, a sampling value is utilized to construct a data segment suitable for online calculation, the EEMD method is utilized to extract network voltage fundamental waves, harmonic interference is filtered out, the sage depth and phase jump angle of a current network voltage are calculated through the two-point method, and finally voltage sag online rapid detection is realized. Therefore, an original signal can be decomposed into IMF with different frequencies, and most harmonic components can be filtered out; and in combination with the EEMD method, the problem that the two-point method is notapplicable under the situation that harmonic distortion is contained is solved, and the advantages of rapidity and accuracy of the two-point method are reserved.

The invention discloses a lightweight pedestrian and vehicle detection method based on improved YOLO v4. The method comprises the following steps: 1) dividing a data set into a training set, a verification set and a test set; 2) performing clustering analysis on a real target frame in the data set label to obtain a priori frame size; 3) building a backbone network by using a Ghost module for extracting visual features of the data set, and reducing parameter quantity while ensuring feature quality; 4) improving the multi-scale sensing capability of the network by using hole convolution with different expansion rates; 5) carrying out feature aggregation on the extracted features; 6) inputting the aggregated features into a detection head, predicting the position and category of the target, and training a model by using a loss function; and 7) inputting a test set image into the network model obtained by training for detection, and outputting a target detection result. According to the method, the pedestrian and vehicle detection precision is guaranteed, meanwhile, the model parameter quantity is small, the detection speed is high, and the requirement for the performance of hardware equipment is lowered.

The invention discloses a power distribution network rapid topology identification method and a system based on K nearest neighbor classification, and a readable storage medium. According to the method, measurement data of node voltage of the power distribution network are used, a K nearest neighbor algorithm is adopted to classify all nodes according to spatial distances between the nodes, then only mutual information values between node pairs in each classification need to be calculated, and then a mutual information topology recognition algorithm is utilized to recover a topology structureof the power distribution network. Wherein the K nearest neighbor classification algorithm is adopted, and the k-nearest neighbor point of each node is obtained by calculating the spatial distance between the nodes in the power distribution network, so that the calculation amount of the mutual information values of the nodes in the topology identification process is reduced. The characteristic ofhigh accuracy of a topology identification algorithm based on mutual information is reserved, and the operation time of topology identification is shortened.

The invention discloses a fatigue test crack monitoring method and device, and the method comprises the steps: building a crack evaluation information base, calculating the length, number, position and form of a crack, and comprehensively evaluating the danger level of the crack according to the information in the crack evaluation information base. According to the fatigue test crack monitoring method and device, crack monitoring is achieved through the visual processing technology, the fatigue damage critical point can be accurately judged, and test piece failure information and data accuracyare more completely reserved; in addition, a crack evaluation information base is established, after the length, the number, the positions and the forms of the cracks are calculated, the danger levels of the cracks are comprehensively evaluated according to information in the crack evaluation information base, the actual harmfulness of the cracks to the components is comprehensively evaluated, and the requirements of actual engineering under various component testing conditions and application scenes are met.

The invention discloses a recording device of vehicle driving behavior information. A device body of the recording device comprises a time recording part which comprises a G time unit for synchronizing with the GPS time; an M time unit for obtaining the time information of a vehicle; a speed recording part which comprises a G speed unit for matching the GPS to obtain the speed information of the vehicle; an M speed unit for obtaining the running speed information by the vehicle; a collision recording part which comprises a collision sensing unit for sensing collision of the vehicle; a judgment and processing part which is connected with all recording parts and receives the information from all recording parts to perform the judgment and processing work; and a record storage device which is connected with the judgment and processing part to store the driving behavior information records. The recording device of the vehicle driving behavior information records a piece of information per 0.1 second in the manner of being 0.75 second faster than the human body response, thus keeping the information with the best judging effect and accuracy for the identification and judgment of vehicle accidents.

This invention relates to a quotient and arithmetical compliment split-screen display method, which tests the quotient and arithmetical compliment, wherein if the value digit number is small, then it is displayed by one screen; if it is large, then it is displayed by split screens. The invention can display the number on one screen due to the split screen function and switches the numbers through keys.

The invention belongs to the technical field of medical apparatus and instruments, and discloses a multi-purpose anesthesia depth monitoring device for the anesthesia department. The multi-purpose anesthesia depth monitoring device for the anesthesia department comprises a brain signal collecting module, a data transmission module, a master control module, a data storage module, a cortex activitycalculating module, a balanced anesthesia depth calculation module, a Fourier transform module and a display module. For the multi-purpose anesthesia depth monitoring device for the anesthesia department, the brain signal collecting module is used for monitoring electroencephalogram signals of a patient in real time, through calculation of a mathematical model, the anesthesia depth index is obtained, the anesthesia degree of the patient can be monitored quantitatively, and thus the clinical risk caused by excessive anesthesia or insufficient anesthesia of the patient is avoided. Meanwhile, themulti-purpose anesthesia depth monitoring device has the advantage that the anesthesia depth monitoring is more accurate under the combined anesthesia of opiates which have the most wide clinical application and inhaled or vein general anesthetics, and the accuracy for the anesthesia depth monitoring under pure inhalation or vein general anesthetics is reserved.

The invention provides a data fusion system and method based on power line carrier communication. The system comprises a main node, a fusion node and an independent node. The independent node is connected with the main node or connected with the main node through the fusion node; and the fusion node is connected with the main node, or is connected with the main node through the independent node orthe fusion node. The independent node is used for collecting first target data of a first preset type, and determining core data of the first target data. The fusion node is used for collecting second target data of a second preset type, receiving core data of the first target data, splitting the second target data, and determining fused data according to the split second target data and the coredata of the first target data. The main node is used for receiving the fused data. Fusion in the data transmission process is realized, the data transmission space is saved, and the data transmissionaccuracy is ensured.

An apparatus and method for loading a radiomap database (DB) and a terminal device are disclosed. The apparatus for loading a radiomap DB includes a moving information calculation unit, and a loading performance unit. The moving information calculation unit calculates the moving information of a terminal based on the difference between a previous location and estimated final location of the terminal as the terminal moves. The loading performance unit loads a radiomap DB, including information about a neighboring node of each reference point, and a radiomap DB, including information about a neighboring node of a reference point, located in a direction in which the terminal moves, based on the moving information calculated by the moving information calculation unit.

The invention provides a molecular marker related to sheep height and application thereof. PCR amplification and sequence analysis are carried out on a Hu sheep CAP2 gene, which finds that a T / C polymorphic site exists at the 309 site of an amplified fragment, a KASPar primer is further used for detecting polymorphic sites of 1346 sheep and establishing a least square model, correlation analysis is carried out on the genotype and the height, and finally that the amplified CAP2 gene fragment can serve as the molecular marker related to the sheep height is determined. The molecular marker can be used for selecting CC homozygous sheep to enter a core group so as to improve the height character of the sheep, and economic benefits can be increased.

The invention provides a method for detecting the carrier frequency shift and the sampling frequency shift in an OFDM system and a circuit thereof. The method for detecting the carrier frequency shift and the sampling frequency shift in the OFDM system comprises the following steps that: related calculations are done on two adjacent code elements on each sub-carrier with a steering signal in order to obtain the related information of each steering signal; the channel frequency response energy of each carrier and each channel serial number are adopted to obtain the information of each channel; the related information of each steering signal, the channel frequency response energy of each sub-carrier and each channel serial number are adopted to obtain the related information of each steering channel; the related information of each steering channel and the information of each channel are adopted to obtain the related detection information of the frequency shift and the sampling shift. The invention combines all the information and the arithmetic elements necessary to the application, thereby decreasing the complexity of the circuit. The method of splitting operation can be also adopted to decrease the difficulty of the multiply operation and the division operation and to keep the accuracy of information.

The invention discloses a method and a device for calculating the response time of search time. The method comprises the following steps: on the basis of a certain search task, independently obtaining the first time point of a first search event on a back-end server, a second time point of the first search event on a front-end server and a third time point of a second search event after the first search event on the front-end server; and adopting the first time point, the second time point and the third time point to calculate the response time point of the second search event on the back-end server. The first search event and the second search event both have the time points on the servers, the accuracy of the absolute time of the servers is retained, meanwhile, the absolute time is kept consistent with a time difference value recorded by the front-end server so as to bring convenience for carrying out correlation analysis on an event according to the time point when the event happens, an analysis result is taken as a basis for analyzing the search behavior of the user, a search result is optimized, and search efficiency is improved.

The invention discloses a molecular marker related to residual feed intake of sheep and application thereof. Sheep ME1 gene is subjected to PCR amplification and sequence analysis, discovery shows that a C / T polymorphic site exists at the 453rd site of an amplified fragment, a KASPar primer pair is further used for detecting the corresponding polymorphic site and establishing a least squares model, genotypes and residual feed intake of 507 Hu sheep are subjected to correlation analysis, and the existence of the molecular marker related to the residual feed intake of the sheep in the amplified ME1 gene segment is determined. In the breeding practice, TT homozygous sheep can be selected into a core group, and the feed consumption and the breeding cost can be reduced by reducing the residual feed intake.

The invention discloses a target object identification method and device, computer equipment and a storage medium, which are applied to the technical field of Internet, and the method comprises the following steps: obtaining a target object for a target service; performing characterization processing on the target object to obtain target object features; based on the target object features, performing feature processing and then determining a first interestingness feature and a second interestingness feature; performing noise addition processing on the first interestingness feature to obtain a first noise addition feature; based on the second interestingness feature and the first noise adding feature, obtaining an interest point classification result corresponding to the target object through a target classification model; and determining an interestingness label of the target object according to the interest point classification result corresponding to the target object. By means of the method, the accuracy of the interest point classification result output by the model is guaranteed on the premise that important features are reserved, so that the accuracy of target object recognition is improved, and protection on any attribute information can be improved by scrambling the features of the target object.

The invention provides a high-precision numerical control phase shifter for a K wave band and a phase shifting method of the high-precision numerical control phase shifter. The method comprises the steps that three-stage phase shifting is divided, a first-stage reflection type phase shifting unit and a second-stage reflection type phase shifting unit are made to carry out 90-degree phase shifting, and then 180-degree phase shifting control is carried out through a 180-degree phase shifting unit; then a binary control code is converted into a control code for accurately controlling three-stage phase shift through a logic control circuit, so that high-precision and continuous phase shift control is realized, the defects of discontinuous K-band phase shift and large phase shift stepping under traditional phase shift are overcome through the arrangement; meanwhile, the phase shift control circuit can be realized by adopting conventional components, so that the cost is reduced, and the working efficiency is improved. According to the method, the cost is further greatly reduced under the condition that the phase shift stability is improved.