67results about How to "Occupies less computing resources" patented technology

The invention discloses a distributed multi-intelligent-agent real time fault detection method based on neighbor associated condition. By converting node models having faults, a new fault detection model is built, fault signals and system condition variables are decoupled, fault detection difficulty is lowered, and only associated condition information of neighbors can be obtained via nodes. According to the distributed multi-intelligent-agent real time fault detection method, by setting fault detection sub-networks and subjecting sub-network models to corresponding linear conversion, node associated condition information can be converted into condition variables in the above fault detection model, and fault detection tasks can be done via the model. The invention provides a new fault detection solution to observe whether faults occur. The invention provides a neighbor associated condition solution for fault detection results. According to the distributed multi-intelligent-agent real time fault detection method, groupment advantages of a multi-intelligent-agent system are fully used, reliability of detection results is increased without system hardware performance improvement, and the distributed multi-intelligent-agent real time fault detection method is high in application value.

The invention discloses a method for regulating image brightness and contrast ratio. The method comprises the steps of receiving an RGB image and regulation parameters, extracting the RGB value before the image is regulated based on the received RGB image, and obtaining the average brightness value based on the RGB value before the image is regulated; substituting the RGB value before the image is regulated and the regulation parameters into a preset relational expression of the RGB value after the image contrast ratio is regulated and the RGB value before the image contrast ratio is regulated, and performing calculation to obtain the RGB value after the image is regulated, wherein the relational expression is obtained by calculating through a formula transforming the RGB to the HSV, a brightness and contrast ratio regulation formula and a formula transforming the HSV to the RGB; performing brightness regulation on the RGB image based on the RGB value after the image is regulated to obtain the RGB image subjected to regulation; outputting the regulated RGB image. The invention further discloses a device for regulating the image brightness and contrast ratio.

The invention discloses a human body bone point tracking method and system in a two-dimensional video stream. The method comprises: a camera obtains a two-dimensional video stream, a foreground picture is obtained by a foreground extraction module, head point and neck point coordinates are obtained by a face detection module, a system determines whether the head point is in a screen or not, and otherwise, the face detection module carries out operation continuously; a human body is divided into a left part ROI and a right part ROI to carry out detection on other key points, and a shoulder point detection module is used for obtaining left shoulder point and right shoulder point coordinates; a hand point detection module obtains left hand point and right hand point coordinates; an elbow point detection module obtains left elbow point and right elbow point coordinates; and then statistics of credibility of all points is carried out and credible points are displayed.

The invention discloses a frequency detection method used for a microgrid system. The frequency detection method includes that firstly data sampling is carried out on microgrid voltage, the number of sampling in each cycle is N, and the time span of the sampling is T; secondly three voltage threshold values are set according to nominal voltage and sampling frequency; thirdly according to the voltage threshold values, the number of voltage cycles within sampling time is obtained in the method of logical judgment, and initial frequency under the three voltage threshold values is respectively obtained; and fourthly the initial frequency is analyzed and judged, frequency with large errors due to wave distortion is eliminated, and then a precise grid frequency value is obtained. The method is simple and convenient to carry out, easy to achieve, and high in precision.

The invention relates to a construction method and application of a lightweight gesture detection convolutional neural network model, and the method comprises the steps: constructing a lightweight gesture detection convolutional neural network framework based on a SquezeNet convolutional neural network framework and an SSD multi-target detection convolutional neural network framework; Acquiring agesture picture and a background picture, and performing image data enhancement and picture synthesis processing on the gesture picture based on the background picture to obtain a gesture data set; And based on the public data set and the gesture data set, training a lightweight gesture detection convolutional neural network framework to obtain a lightweight gesture detection convolutional neuralnetwork model. According to the invention, a small amount of gesture data is expanded into the gesture data set containing a large amount of picture data at a high speed; The technical problem that alarge amount of high-quality gesture picture data is difficult to obtain is solved, in addition, by combining the SquezeNet convolutional neural network architecture and the SSD multi-target detectionconvolutional neural network architecture, the constructed lightweight gesture detection convolutional neural network model occupies few computing resources, and can be applied to various detection platforms.

The invention belongs to the field, particularly relates to a catheter detection method for a vascular aneurysm interventional operation navigation system, and aims to solve the problem that an interventional catheter in an X-ray transmission image in an intravascular aneurysm repair operation cannot be accurately segmented and tracked in real time. The method comprises the following steps: acquiring an X-ray transmission video sequence of a region containing a catheter as a to-be-detected sequence, generating a segmentation mask sequence through a trained encoding and decoding structure according to the to-be-detected sequence, and covering the to-be-detected video with a binary segmentation mask sequence to obtain a video sequence of the catheter. The accuracy and speed of segmentation and tracking of the X-ray transmission image interventional catheter in the intravascular aneurysm repair operation are improved, and the real-time performance required by an operation tracking algorithm is met.

The invention provides a multi-target tracking method based on the theory of random sets, and aims at realizing multi-target state effective tracking in a complex scene under the restricted condition of limited resources. According to the method, firstly each continue live track is further predicted according to the Bayes rule in the track prediction stage, and then a target track is self-adaptively generated according to target prior information; and in the track updating stage, firstly the corresponding existence weight probability and a combined multi-target probability density function are computed under each assumption, and then the posterior probability parameter of each target mark number is computed, including the existence probability and the corresponding probability density function. The method has the advantages of being low in approximation cost and excellent in performance under the limited resources and has the robustness suitable for any measurement models. Besides, the algorithm occupies less system computing resources so as to have great engineering application prospect.

The embodiment of the invention discloses a method for suppressing distance ambiguity based on antenna pattern synthesis. The method comprises the steps of obtaining a target wave position working instruction; determining a target wave position signal parameter according to the target wave position working instruction; initializing measurement parameters which comprise the target wave position signal parameter, an SAR system parameter and a user-defined parameter; determining a signal emission weight corresponding to the target wave position signal according to the measurement parameter; determining a signal correlation matrix and a range ambiguity correlation matrix of the target wave position signal according to the signal emission weight and the measurement parameters; determining an initial array weight according to the signal correlation matrix and the distance ambiguity correlation matrix; determining a convex optimization problem according to the target wave position working instruction; and determining a receiving weight corresponding to the target wave position signal according to the initial array weight and the convex optimization problem. The embodiment of the invention also discloses a device for suppressing distance ambiguity based on antenna pattern synthesis.

The invention discloses a method for regulating the image saturation degree. The method comprises the following steps that RGB (red green blue) images are received, parameters are regulated, and the RGB value before the image saturation degree regulation is extracted on the basis of the received RGB image; the RGB value before the image saturation degree regulation and the regulation parameters are substituted into the relational expression of the RGB value after the image saturation degree regulation and the RGB value before the image saturation degree regulation, and the RGB value after the image saturation degree regulation is obtained through calculation, wherein the relational expression is obtained through calculation by an RGB to HIS (hue saturation intensity) formula, a saturation regulation formula and an HIS to RGB formula; the RGB image is subjected to saturation degree regulation on the basis of the RGB value after the image saturation degree regulation, and an RGB image after the regulation is obtained; the RGB image after the regulation is output. The invention also discloses a device for regulating the image saturation degree.

The invention discloses a variable-speed scheduling method for mixed transportation of crude oil in a long-distance pipeline. The variable-speed scheduling method aims at the problems that a traditional manual production scheduling scheme in the long-distance pipeline mixed oil transportation process is low in efficiency and large in time delay of the long-distance pipeline oil transportation, theoriginal established production scheduling scheme cannot be used during transporting oil with variable speed according to the actual production situation, and proposes a variable-speed production scheduling system with self-adapting variable-speed oil transportation of mixed transportation of crude oil in the long-distance pipeline, and the production scheduling scheme can be independently generated after user configuration without manual intervention. Compared with a commonly used mixed integer programming method, the method is small in calculation scale, high in solving speed, low in probability of no solution and less in computing resource occupation; at the same time, according to the method, the problem that the original production scheduling scheme cannot be used after changing theoil transportation speed in case of the long-distance pipeline existing in the oil transportation process is solved; and the production efficiency of enterprises is improved, the loss of the enterprises is lowered, and the economic benefit is improved.

The invention discloses a method and a system for obtaining a noise horizontal curve of an image sensor, and relates to the technical field of image processing. The invention aims at providing the noise horizontal curve estimation method and system with the advantages that the calculation is simple, and the accuracy degree is high. The method is characterized in that a color image is subjected to single-channel grayscale image decomposition to obtain grayscale images with different color channels; the three grayscale images are sequentially processed to obtain the sensor noise horizontal curve of each grayscale image; the grey level histogram statistics are carried out; the trusted grayscale interval proportion factor of the Trusted gray is calculated; the trusted grayscale interval is calibrated; the grayscale images are blocked; the variance and the mean value of the trusted image block are obtained; the mean value and the variance of an end point are obtained; each mean value and each variance are used as points on the curve, and the noise horizontal curve of the grayscale image sensor is obtained by a segmented three-time Hermite interpolation algorithm.

The invention discloses an adaptive control method for a spectrograph applicable to a cold environment and relates to an automatic control technology. The adaptive control method comprises the following steps: 1) roughly measuring the intensity of incident light; 2) determining an overall output maximum value of the spectrograph; 3) decomposing overall output of the spectrograph; 4) determining the best exposure time top of the spectrograph; and 5) carrying out adaptive measurement on the spectrograph, setting the exposure time of the spectrograph as top, and carrying out measurement. The adaptive control method disclosed by the invention has the beneficial effects that by virtue of a way of estimating the intensity of the incident light and automatically regulating the exposure time, the spectrograph can be automatically adaptive to variation of external environment temperature and light field intensity, especially in a severe cold environment, the adaptive measurement is carried out, and measurement errors, caused by temperature variation, of devices are reduced, so that a satisfying measurement result is obtained.

The invention discloses an image encoding, recognizing and tracking method, system and device and a storage medium. The image encoding method comprises: selecting N target graphs; performing encodingat a set position of a connecting line of any two target graphs to generate encoding information corresponding to the connecting line; and generating a target image including the encoding informationand the N target graphs. According to the method, the robot can uniquely identify the target image from one frame of image, and even if other obstacles are shielded in the identification process, following is not influenced; once the target object is lost, the target object can be quickly identified again, the condition of tracking the target object by mistake is avoided, and the method has the advantages of high calculation efficiency, high identification speed, timely response, low cost, no interference from the environment, capability of realizing tracking anytime and anywhere, no information leakage and the like.

The invention discloses an OFDM channel estimation device and method. The OFDM channel estimation device comprises an amplitude value comparison module, a loop filtering module, a predicated value generation module and an estimation value generation module. One input end of the amplitude value comparison module is used for receiving an observation value, the other input end of the amplitude value comparison module is connected with the output end of the predicated value generation module, the output end of the amplitude value comparison module is connected with the input end of the loop filtering module, the output end of the loop filtering module is connected with the input end of the predicated value generation module, and when the error value output by the amplitude value comparison module is within a preset error range, an estimation value output by the estimation value generation module is the final result. The channel estimation device is clear in structure and easy to obtain, a few computing resources are occupied, and the performance of the channel estimation device is better than that of a Kalman filter using a first-order autoregressive model through a traditional method.

The invention relates to a few sample classification method based on attention model, which comprises the steps of training a convolutional neural network image classification model, removing a full connection layer of the convolutional neural network image classification model after the convolutional neural network image classification model is trained, and reserving a convolutional neural network part of the convolutional neural network image classification model; performing basic classification on the to-be-detected image according to the feature vectors of all the sample images to obtain n* m scalar quantities, n being the number of classes of all the sample images, and m being the number of sample images in each class; connecting m sample images of each type into an m * M-dimensionalvector, and inputting a nonlinear mapping function to obtain n * m relative weights; multiplying the obtained n * m scalar quantities by the corresponding n * m relative weights to obtain new n * m scalar quantities, and adding the new m scalar quantities in each class to obtain a final classification result. According to the method, by mining the score of the importance of the intra-class images, the model can pay more attention to important samples, and the intra-class images are balanced.

The invention discloses a pedestrian re-identification method based on a TOIM loss function, and the method comprises the following steps: S1, preparing a data set, and enabling pedestrians with different identities in the data set to have different IDs; S2, establishing a network model based on a ResNet-50 model by utilizing a pytorch framework; S3, inputting the pictures of the training set in the data set into the network model for training; and S4, inputting the pictures of the test set in the data set into the trained network model, and identifying the pedestrian identity. According to the pedestrian re-identification method, the advantages of an online instance matching (OIM) loss function and a triple (Triplet) loss function are combined, meanwhile, attention to difficult samples isemphasized, and the batch processing construction process of the triple loss function is simplified, so that the convergence speed is greatly increased, and the pedestrian re-identification accuracyis effectively improved.

The invention provides a coal distribution detection method and device, and the method comprises the steps: recognizing a plurality of pieces of raw coal, and obtaining the coal quality information corresponding to each piece of raw coal; obtaining to-be-paired labels with the same number as the multiple pieces of coal quality information from the bidding machine, wherein unique number informationis stored in each to-be-paired label; paring the multiple to-be-paired labels with the multiple pieces of coal quality information in a one-to-one mode, writing the coal quality information into thepaired to-be-paired labels to obtain multiple coal quality information labels, wherein the coal quality information labels are attached to the corresponding raw coal; and reading the coal quality information labels on a plurality of raw coals entering the raw coal bunker through reader-writers arranged in the raw coal bunker at intervals to obtain a plurality of pieces of serial number information. The RFID technology can be applied to the raw coal bunker, a scientific reader-writer arrangement method is provided, the coal distribution condition is precisely predicted in the raw coal bunker, the coal quality of the coal before entering the boiler for combustion is accurately speculated, and combustion of the boiler is assisted to be regulated and controlled.

The invention discloses a power distribution network electric energy distributed transaction model based on distributed database support, which is characterized in that electricity purchasing and selling parties send electricity purchasing and selling information through a distributed database network, and an electric energy transaction negotiation market chain code model completes negotiation oftransaction electric quantity and transaction price to form a temporary contract; the network security constraint chain code model performs security check on the temporary contract line power flow andthe node voltage; and the on-chain and off-chain model determines the temporary contract passing the security check as a final contract, and executes a transaction. According to the invention, the market main body is allowed to participate in the power market in a decentralized manner and negotiate the transaction electric quantity and the transaction price, the electricity purchasing cost of themarket main body can be effectively reduced, and the electricity selling income of the market main body is increased. Line power flow blocking and node voltage deviation conditions can be effectivelyjudged, and safe operation of a power distribution network is facilitated. The constructed distributed database capacity expansion architecture can ensure the safety and traceability of all historical data storage while ensuring the computing power of the distributed database.

The invention discloses a frequency detection method used for a microgrid system. The frequency detection method includes that firstly data sampling is carried out on microgrid voltage, the number of sampling in each cycle is N, and the time span of the sampling is T; secondly three voltage threshold values are set according to nominal voltage and sampling frequency; thirdly according to the voltage threshold values, the number of voltage cycles within sampling time is obtained in the method of logical judgment, and initial frequency under the three voltage threshold values is respectively obtained; and fourthly the initial frequency is analyzed and judged, frequency with large errors due to wave distortion is eliminated, and then a precise grid frequency value is obtained. The method is simple and convenient to carry out, easy to achieve, and high in precision.

The invention provides a filtering method and device and a storage medium, and relates to the technical field of slow signal filtering. According to the filtering method, firstly, collected signal values from i-m-th time to the i-m+n-th time are acquired, and an original signal value of i-m-1-th time is acquired; a judgment result that the original signal is, from the i-m-th time to the i-m+n-th time, in a rising trend, a falling trend or status of remaining unchanged is acquired on the basis of the collected signal values from the i-m-th time to the i-m+n-th time and an original signal valueof i-1 time; then correcting an collected signal value of i-th time on the basis of a change estimation equation corresponding to the rising trend, the falling trend and the status of remaining unchanged; and thus an original signal value of the i-th time after filtering is obtained. According to the filtering method, estimation on slow signals is realized through simple numerical comparison and addition and subtraction operations, operation speed is higher, occupied resources are fewer, and the original slow signals can be quickly and stably extracted from signals under interference.

The invention discloses an imbedded voice synthesis method based on an adaptive weighted spectrum interpolation coefficient, which is used in an imbedded operation system for transforming arbitrary received characters into voices and outputting the voices. The method comprises the following steps of: at a training end, extracting a base frequency adaptive weighted spectrum interpolation (a STRAIGHT spectrum) for voice signals, extracting a track spectrum characteristic coefficient for the STRAIGHT spectrum, and modeling and training the characteristic coefficient through HTS; and at a synthesis end, after a characteristic coefficient sequence is calculated via a model, acquiring a synthesis voice by a traditional parameter synthesizer. When the method provided by the invention is used, the synthesis voice quality equivalent to that of a STRAIGHT synthesizer can be acquired, and the STRAIGHT synthesizer is replaced by the traditional parameter synthesizer at the synthesis end so as to greatly improve the synthesis speed, and the embedded application becomes possible.

The invention discloses a parking lot ground Chinese sign recognition method and system and a storage medium, and the method comprises the steps: obtaining a parking space image, and carrying out thepreprocessing; performing Chinese character recognition on the preprocessed image by using a neural network model; when the Chinese character does not exist in the image, enabling the neural network model to output an identification result without the Chinese character; when the Chinese characters exist in the image, enabling the neural network model to extract a text region image containing the Chinese characters and outputsthe text region image; when the neural network model outputs a recognition result without Chinese characters, ending the process; and when the neural network model outputsa text region image, performing text correction on the text region image, performing character segmentation on the corrected text region image to obtain a plurality of character region images, classifying the plurality of character region images, and outputting an identification result of the ground Chinese sign. According to the invention, the ground Chinese sign of the parking lot can be identified, and the vehicle is prevented from being parked in an improper area during parking.

The invention relates to a calculation method for intelligent auxiliary driving lane changing early warning of an automobile. The calculation method comprises the following steps: a, setting a preset safety distance in an automobile intelligent auxiliary driving lane changing system; b, when the main vehicle changes the lane or is about to change the lane, estimating the distance between the front vehicle and the rear vehicle on the target lane; and c, when the distance between the front vehicle and the rear vehicle on the estimated target lane is smaller than the preset safe distance, sending out warning information, and meanwhile forbidding lane changing operation; and when the distance between the front vehicle and the rear vehicle on the estimated target lane is greater than a set safety distance, carrying out lane changing operation. According to the calculation method for intelligent auxiliary driving lane changing early warning of the automobile, occupied calculation resources are low, it can be guaranteed that data are updated in real time in the automobile speed changing process, safety can be improved, real-time performance can be guaranteed, meanwhile the calculation method can be combined with other intelligent auxiliary driving technologies to form a more intelligent and effective function, and user experience is enhanced.

The invention discloses a vehicle driving control method and device, and equipment, and relates to the field of vehicle control, and the method comprises the steps: obtaining the current first driving state data of a vehicle and the second driving state data of a target vehicle; determining a current vehicle distance between the vehicle and the target vehicle and a target vehicle distance between the vehicle and the target vehicle; based on a preset acceleration planning model, determining an acceleration control strategy of the vehicle according to the first driving state data, the second driving state data, the current vehicle distance and the target vehicle distance, and determining the acceleration planning model based on a piecewise function; and controlling the driving state of the vehicle according to the acceleration control strategy. Acceleration planning results under different working conditions can be simply and quickly calculated, the calculation amount is small, the occupied calculation resources are few, implementation is easy, meanwhile, the acceleration planning results accord with human driving habits, and the comfort level of drivers and passengers can be effectively improved.

The invention provides a satellite navigation high-precision relative positioning method based on ranging information assistance. The method comprises the following steps of selecting a primary satellite, establishing a satellite navigation double-difference observation equation, acquiring ranging information, establishing a ranging information observation equation, solving a baseline vector after an ambiguity floating solution and a fixed solution and completing satellite navigation relative solution. According to the method, the ranging information is introduced, strong constraint information of the distance between the base station and the mobile station is provided for satellite navigation relative positioning, and the reliability of satellite navigation relative positioning can still be guaranteed by using the information when the number of satellites is insufficient or the geometric configuration is poor. The method provided by the invention is simple in algorithm model, has no obvious increase in operand compared with the original algorithm, and occupies less operation resources. A ranging information error model is simple, complex modeling is not needed, and implementation is easy.

The invention discloses a food cooking degree identification method, device and system, an electric appliance, a server and a medium. The food cooking degree identification method comprises the following steps: acquiring an initial state image of food; obtaining a target image according to the initial state image and a preset adversarial neural network model, the target image representing an image when the food reaches the required cooking degree; acquiring a current state image of the food; and comparing the similarity between the target image and the current state image to obtain the food cooking degree. According to the food cooking degree recognition method, the cooking degree of the food is obtained by comparing the similarity of the target image and the current state image, the requirement for cooking degree judgment of various types of food can be met, the generalization degree is high, in addition, the method occupies small computing resources, and the cooking degree judgment cost is reduced.

The invention discloses a vehicle driving control method, device and equipment, and relates to the field of vehicle control, and the method comprises the steps: determining a current distance and a target distance between a vehicle and a target vehicle; determining current first driving data of the vehicle and second driving data of the target vehicle; inputting the first driving data, the second driving data, the current distance and the target distance into a preset control planning model, determining an acceleration control strategy of the vehicle, and determining the control planning model is determined based on a polynomial function; and executing the acceleration control strategy to control the driving of the vehicle. Acceleration planning results under different working conditions can be simply and quickly calculated, the calculation amount is small, the occupied calculation resources are few, implementation is easy, meanwhile, the acceleration planning results accord with human driving habits, and the comfort level of drivers and passengers can be effectively improved.

The invention discloses an image encryption method. The method comprises the following steps: obtaining a to-be-encrypted image, a first chaotic mapping matrix and a second chaotic mapping matrix; multiplying the to-be-encrypted image by the first chaotic mapping matrix to obtain a first image; carrying out discrete cosine transform on the first image and the second chaotic mapping matrix to obtain a second image and a third matrix; multiplying the third matrix by the second image to obtain a third image; and performing inverse discrete cosine transform on the third image to obtain an encrypted image. Thus, when the method is adopted to encrypt the image, the occupied computing resources can be reduced, and the storage space can be reduced, so that the method can be suitable for equipmentwith poor computing performance and / or small storage space, and due to the fact that the occupied computing resources are small, computing is simple, and the real-time requirement of image encryptionprocessing can be met. The invention further discloses an image encryption device and equipment and a readable storage medium, which have corresponding technical effects.

The invention discloses an autonomous learning sonar target individual recognition method, which comprises the following steps of: constructing a deep learning network model for sonar target individual recognition, training the model by using historical data to obtain an initial model and a template library, then receiving new data, predicting the new data by using the initial model, extracting an individual feature template, and finally extracting an individual feature template; the method comprises the following steps: performing retrieval matching in a template library, realizing rapid individual identification or identity determination of a new target, updating the template library, selecting a model autonomous learning updating method according to sonar platform characteristics, selecting an updating method based on total historical data joint learning if the platform is a platform with sufficient hardware resources, and selecting an updating method based on total historical data joint learning if the platform is a platform with limited hardware. And if not, selecting an updating method based on field data incremental learning, and continuously learning new data to improve the model performance and better meet actual combat environment requirements.

The invention discloses a multi-means and low-cost fan fault monitoring method, which includes the following steps: (1) defining a fan fault state set according to the common faults of fans in practical application; (2) optimizing a fault feature set by using a common fan fault feature set template and the existing fault data of a target fan; (3) continuously collecting the instantaneous current I, surface temperature T and vibration L signals of the fan, and calculating the feature values of the current, temperature and vibration signal of the fan at the current time; and (4) matching the calculated feature values of the current, temperature and vibration signal of the fan with the items in the fan fault feature set, and judging that the current fault state of the fan is the fault state represented by the matching item in the fault state set if the calculated feature values of the current, temperature and vibration signal of the fan are consistent with the items in the fan fault feature set. By monitoring and analyzing the current, temperature and vibration signals of fans, timely and accurate monitoring of fan faults is realized. The multi-means and low-cost fan fault monitoringmethod has the characteristics such as low cost, easy integration and low difficulty in use.