307results about How to "Reduce the amount of data calculation" patented technology

The invention relates to a vibration monitoring structure based on optical fiber polarized light time domain reflection sense, comprising a laser (1), a first coupler (2), a second coupler (7), a third coupler (8), a fourth coupler (11), a fifth coupler (13), an acoustic optical modulator (3), a polarizer (4), a circulator (5), a sensing optical fiber (6), a first analyzer (9), a second analyzer (10), a first detector (12), a second detector (14), a first low-pass filter (15), a second low-pass filter (16), a first analog-to-digital converter (17), a second analog-to-digital converter (18), anFPGA (19) and a microprocessor MCU (20); the method comprises the following steps: measuring the changes of the polarization state of each optical fiber section with time according to a plurality ofmeasurements, combining Fourier transformation technology to obtain frequency spectrum information, and analyzing the frequency spectrum information to obtain vibration information. Compared with theprevious method which is adopted by a POTDR system and carries out iteration operation on polarization state, the method has the advantages of simple algorithm and small data operation and greatly reduces monitoring time and cost of the system.

The invention provides a path planning method and system for a mobile robot. The method includes: establishing a map according to the traveling environment data of the mobile robot; dividing the map into a plurality of uniform grids to obtain an initial grid map, and determining the initial position and the final position of the mobile robot on the initial grid map; receiving an obstacle signal detected by a sensor on the mobile robot; determining the repulsive force of the mobile robot at the current position according to the obstacle signal; according to the repulsive force of the mobile robot at the current position, constructing an attenuated repulsion field of an obstacle area; assigning values to respective grids in the initial grid map according to the attenuated repulsion field to obtain an assigned grid map; and generating a traveling path of the mobile robot toward the final position by using a non-traversal search approach. The method and system do not involve the vector calculation of potential field force at all when planning an optimal path and eliminates the possibility of deadlock of a local optimum point that may occur in an artificial potential field method.

The invention provides an H.264 video encryption method based on a region of interest. According to the method, an encryption processing scheme based on a three-level secret key is adopted, after a video encoding parameter of the region of interest in an H.264 video is changed by means of the three-level secret key, and then encoding processing is carried out by the direct adoption of an H.264 video encoding criterion, so that the effect of partial encryption of the region of interest is achieved, the situation that a pixel value in the region of interest is directly processed for encryption is avoided, data computation amount and operation complexity are reduced, encryption processing operation speed is beneficially increased, the increasing amount of the code rate after encryption is reduced, encrypted video data can be compatible with an H.264 standard encoder in a data format and can be directly applied to real-time playing of the H.264 video, besides, the method can eliminate encryption diffusion influences of encryption of the region of interest on a region of non-interest by combining and using a motion estimation drifting elimination technology, and the reading effect of the encrypted videos is improved.

The invention discloses a face recognition system, which relates to the technical field of image processing and pattern recognition. First, the gender of a face image to be recognized is determined through a deep learning method. Then, the face identity of the face image to be recognized is determined with high accuracy according to the gender identified through deep learning and based on a convolutional neural network. Finally, the facial micro expression of the face image to be recognized is determined through a binocular vision technology based on the parallax characteristic of the face image to be recognized. Face recognition based on the gender information, identity information and facial expression information of a face image to be recognized is realized. By combining the deep learning method, the convolutional neural network and the binocular vision technology, the influence of factors such as illumination and facial gesture on the face recognition rate is reduced, the amount of computation in the face recognition process is reduced, fast and high-accuracy face recognition is realized, the cost of face recognition is reduced, and the accuracy of face recognition under different facial postures is improved.

The invention provides a yawning action detection method for detecting fatigue driving. The yawning action detection method comprises the following steps: relative position relations of facial feature regions in a facial image region in a video image are respectively determined in a matched manner by virtue of feature region contours respectively corresponding to the facial feature regions of a face matching template, so as to well ensure the accuracy of mouth location, quick matching location of the mouth regions in the facial image region in the video image is carried out by adopting an active shape model matching algorithm, the data operation amount is small, the processing speed is high, and the real-time performance of the mouth location is ensured; then, the actual shapes of mouth feature region counters are determined by carrying out matching location on the mouth regions in the facial image region in the video image to recognize the mouth opening or closing state. The yawning action detection method realizes the detection of the yawning action, is high in detection accuracy, fast in speed, provides an effective and high-real-time-performance solution for the detection of the yawning action, and can provide an alerting signal with timeliness for the fatigue driving detection.

The invention discloses a text similarity obtaining method and a device. The text similarity obtaining method and the device can accurately and rapidly calculate similarity of texts, operation is simple, and data calculation amount is small. The text similarity obtaining method comprises the following steps. Stop words in all texts are removed according to scheduled stop word rules, and available words of the texts are extracted. Hash values of all the available words in the texts are calculated, and similar hash values of the text are obtained according to the hash values of all the available words in the texts. Similarity among different texts is obtained by utilizing the similar hash values of the texts.

The invention relates to the technical field of the traffic illegal detection and in particular to a traffic illegal detection method based on vehicle tracking and a system. The method comprises the following steps: pre-building a vehicle 3D model according to monitoring video data of an actual road; and in the process of actual detection, marking and tracking the vehicle by using the vehicle 3D model, judging whether the vehicle has the traffic illegal act according to the video image data generated by the tracked vehicle. The method is capable of marking the vehicle by using the vehicle 3D model, and using the tracking mode to obtain the video monitoring data of the vehicle. Compared with the mass center and real line distance method of abundant foreground change, the method is capable of eliminating the interferences, such as vehicle tracking, blocking, pedestrian and non-motor vehicle, and improving the accuracy rate of traffic illegal act judgment.

The invention discloses an optic axis and inertia axis superposition-based geographic tracking method which comprises the steps of computing the space pose of a carrier aiming line in real time by using a space two-point geometric algorithm according to the positioning data and the self position information of a target, computing the space angle of the aiming line relative to a carrier coordinate system by being combined with the carrier space pose and the space coordinate transformation principle, sending the angle to a servo control unit, and adjusting corresponding space pose of a turntable by the servo control unit so as to complete the geographic tracking to the target. The method is different from the image tracking method in terms of theory, therefore, the method is not influenced by the cloud, the mist and the obstacles; and when being used on aircrafts, the method is capable of avoiding the problem of losing the target due to strenuous vibration or sight shielding, so that the scouting and searching capabilities of the battlefield weapon systems are remarkably improved. The method can also be used for guide devices of various guided missiles, so as to improve the accurate attacking capability to ground targets.

A phase perturbation based island detection system for a distributed gird-connected inverter is disclosed. The island detection system comprises a voltage controller, a multiplying unit, a summing unit, a current controller and a pulse width modulation generator which are serially connected in sequence, wherein the input ends of the voltage controller, the multiplying unit, the summing unit and the current controller are connected with direct current bus voltage and set direct current bus standard voltage respectively; the pulse width modulation generator is for outputting four paths of triggering pulse signals used for driving a single-phase H bridge inverter of a single-phase distributed gird-connected power generation system; the island detection system also comprises an orthogonal signal generator, a synchronous phase-locking device, a phase perturbation unit and a threshold value judging unit which are serially connected in sequence, wherein the input end of the orthogonal signal generator is connected with the voltage end of a point of common coupling of the single-phase distributed gird-connected power generation system; and the output end of the phase perturbation unit is connected with the input end of the multiplying unit. The phase perturbation based island detection system for the distributed gird-connected inverter has rapidness in island detection, stability of the detection system, and possibility of being used as island detection standard, can satisfy demand on percent of pass of the power quality, and has the important and actual application value in safe operation of the distributed new energy gird-connected power generation system.

The invention relates to a method for establishing a public transport transfer network, which comprises the following steps of: receiving data, namely receiving position coordinates of a starting point and a terminating point; generating a transfer network, namely setting bus lines between the starting point and the terminating point as transfer network nodes, and connecting two nodes of any two bus lines with a transfer relation to establish the transfer network from the starting point to the terminating point; and generating transfer paths, namely selecting at least one transfer path from the transfer network as an output result. The invention also relates to a device for establishing the public transport transfer network. The method and the device have the advantages of greatly reducing the data calculation quantity for acquiring the transfer paths by setting the bus lines as calculated nodes of the transfer network, and quickly outputting the paths of the public transport transfer network.

The invention discloses an image recognition technology-based forest fire automatic monitoring and recognition system and method, and belongs to the field of image processing. Based on the characteristics of wide application range and high economy of video monitoring image processing and mode recognition technologies, the image recognition technology-based forest fire automatic monitoring and recognition system and method are applied to the field of forest fire monitoring and early warning. The image recognition technology-based forest fire automatic monitoring and recognition method comprisesthe following steps: performing preprocessing of image noise reduction and data compression by using median filtering and gray processing methods; detecting a moving target in the monitoring image byusing a Gaussian mixture background model and a differential detection technology; extracting multi-dimensional features such as color moment, energy, entropy, inverse difference moment and circularity from the detected target image, and fusing the multi-dimensional features into a normalized feature vector; and finally, using a binary classifier of a support vector machine for identification andjudgment. According to the image recognition technology-based forest fire automatic monitoring and recognition system and method, a moving target appearing in a complex background can be well detected, and characteristic parameters are extracted for identification, and automatic fire identification is further realized, and the image recognition technology-based forest fire automatic monitoring and recognition system and method have the advantage of high monitoring accuracy.

The invention discloses a method and system for identifying an oil spilling target of a ship. The method comprises the following steps of: establishing a conventional oil type actual measurement spectrum database of the ship by use of actual measurement spectrum data of a conventional oil type through an ASD (Analytical Spectral Device) ground spectrometer; synchronously acquiring actual measurement spectrum data and imaging spectrum image data of different oil types by utilizing the ASD ground spectrometer and a vehicle-mounted high-spectrum sensor, acquiring rebuilt spectrum data by utilizing an experimental linear method, and carrying out precision evaluation on the rebuilt spectrum data by utilizing a similar coefficient method and optimizing a rebuilt scheme; and optimizing spectrum characteristics of an oil film according the actual measurement spectrum data and the rebuilt spectrum data, and selecting a spectrum section to carry out spectrum angle map identification. According to the invention, a problem that continuous spectrum data on the same ground can not match because absorption characteristics of multi-waveband images and satellite-mounted high-spectrum images are attenuated due to space scale change is solved, the oil spilling identification precision is improved to more than 85%, and fast identification of the oil spilling target based on a high-spectrum image is realized.

The invention discloses an image searching method and device. The method comprises steps of acquiring at least one target area selected from an input image by the user, extracting image features of the target area, searching an image database according to the image features of the target area, acquiring an image of an area matched with the image features of the target area and designating the image as a search result. The image searching method provided allows a user to select a local area on the image during image searching, and retrieve an image database according to features of the local area to acquire a searching result, so comparison is only required between the features of the local area and image features of the image database; data computation amount can be improved; besides, an image local search function can be achieved; and a demand for user searching image local areas can be met.

The invention provides a named entity recognition method based on a neural network and a computer storage medium, and the method comprises the steps: inputting a to-be-recognized character string intoa classification model, recognizing the language intention category of the character string through the classification model, and searching an entity label set corresponding to the recognized language intention category from a preset mapping table; inputting the character string into a named entity model to sequentially identify each character in the character string to obtain probability valuesof a plurality of entity tags to which the words belong in the character string; searching an entity label matched with the entity label of the word contained in the character string from an entity label set corresponding to the language intention category, and aiming at the matched entity label, selecting the entity label of which the probability value ranks top N in the matched entity label setas the entity label of the corresponding character. The incorrect entity tags in the named entity model recognition result are filtered through the language intention category recognition result of the classification model, so that the error recognition rate of the named entity model is reduced.

The invention discloses a method and a device for realizing a three-dimensional rotating interface based on the Android system. The method comprises the steps as follows: feature information of a to-be-realized three-dimensional interface of a target object is extracted, a three-dimensional coordinate system is selected according to the extracted three-dimensional interface feature information, and a mathematical model mapping the three-dimensional interface feature information is constructed; the target object is loaded to the mathematical model; the mathematical model on which texture mapping is performed is subjected to coordinate transformation, and an inclined mathematical model matched with the feature of the to-be-realized three-dimensional interface is acquired; the inclined mathematical model is triggered to rotate, so that a rotating mathematical model is obtained; the rotating mathematical model is subjected to perspective projection according to a preset viewpoint, so that sight plane coordinate information corresponding to the rotating mathematical model is obtained; and the sight plane coordinate information is converted into screen plane coordinate information, and the three-dimensional rotating interface is presented according to the converted screen plane coordinate information. With the application of the method and the device, the interface display effect can be improved, and the computational cost for realizing the three-dimensional rotating interface is reduced.

The invention discloses a network-on-chip congestion control method based on a dynamic routing table, and belongs to the field of network-on-chip congestion control methods. The method comprises the following steps: defining the self-updating rule of the dynamic routing table; starting and initializing a network-on-chip; resolving an own dynamic routing table; determining non-fault adjacent nodes; and delivering data. According to the method, the dynamic routing table is adopted on the network-on-chip for the first time, the own current optical transmission path of the dynamic routing table can be dynamically updated and decided according to the current states of other nodes, and global awareness can be better shown; and optimal paths with smallest global congestion degrees to any other destination node are always saved in each dynamic routing table, so that convenience is brought to rapid reading and use, and the repeated operation time is shortened greatly. Moreover, through the method, frequent reading and writing operation among routing nodes can be reduced, the data computation amount is reduced, and the operation time is saved, so that network congestion is relieved and limited, and the data processing capability and throughput of the network-on-chip are increased.

The invention discloses an adaptive neural network controller of a bridge crane. A dynamic model of a bridge crane is established; linear processing is carried out on the dynamic model of the bridge crane; an external disturbance factor compensation item d is introduced to obtain a linear model of the bridge crane; and an adaptive neural network controller including a trolley controller and a loadcontroller is obtained based on a RBF neural network design. According to the invention, on the basis of the neural network adaptive method, the double-feedback adaptive controller is designed for trolley positioning and load oscillation prevention of the crane; and on the basis of the self-learning method, random approximation of non-linear parts like the model error and the external interference during the crane modeling process is realized, so that stability control is realized.

The invention belongs to the technical field of voice evidence obtaining. The invention discloses mobile device source identification method and system based on multimode fusion depth features. The method comprises the following steps: firstly, extracting MFCCs and GSV features of test data, correspondingly segmenting the features into multiple paths, then respectively training CNNs and performingfusion to obtain fused depth features, then determining the fused depth features by using a trained depth residual error network, and finally carrying out joint decision on the determination resultsof short samples of each path by adopting a voting method. According to the method, when the GMM-UBM model is trained, the data is screened according to the characteristics of phonemes and tones of the voice data, and a small amount of representative data is selected, so that the representation generalization of the model is ensured, the data calculation amount is reduced, and the modeling efficiency is improved; according to the method, the deep neural network is used for supervised training to extract the deep features, redundant and interference information in the feature data is eliminated, the feature data is simplified, the characterization of the data is improved, the dimensionality of the data is reduced, and the calculation amount is simplified.

The present invention provides an adversarial transfer learning method and system for cross-border big data analysis. The method includes: substituting the random multi-linear fusion representation corresponding to each tensor in the tensor set corresponding to the unlabeled data set in the source domain and the target domain into the original loss function of the discriminator to obtain the current loss function of the discriminator, and Using backpropagation, adjust the parameters of the discriminator to minimize the current loss function as the current best loss function of the discriminator; the tensor in the tensor set is in the preset deep neural network The tensor product of the data vectors of all data layers in the preset data layer set; update the preset deep neural network parameters based on the current best loss function and enter the next update of the preset deep neural network parameters until the parameters converge . The joint distribution offset of multiple data layers in the preset deep neural network preset data layer set obtained by the present invention is reduced, and the application effect is better in the target field.

The invention relates to the technical field of information security, and the embodiment discloses an SM2-based both side signature method and system. By adoption of the technical scheme provided by the invention, a part of private key can be separately stored in both communication sides, a signature or decryption and other operations can only be performed on a message by the cooperation of the both sides, neither of the both communication sides can obtain any information of the private key of the other side, therefore an attacker cannot forge the signature or decrypt the ciphertext if invading any of the both sides, and the both communication sides need to transmit two data to the opposite side in the signature interaction, so that the data communication traffic and the data calculation amount are low, and the signature efficiency is higher.

Disclosed is a spectral calibration method for a swept source optical coherence tomography (SS-OCT) system. A transparent sheet is placed in a reference arm of the SS-OCT system. A fixed optical path difference interference signal having an optical path difference of 2nd can be obtained by using the transparent sheet. Interference signals of the system are collected by dense sampling and contain interference signals of the sheet and sample interference signals, the spectral calibration is performed using the interference signals of the sheet, the sample interference signals at equal intervals of the wave number are obtained, the inverse Fourier transform data processing and image reconstruction algorithm are then carried out, and an tomography structure image of the sample can be obtained. The method of the invention realizes the real-time calibration of the system interference signals by using the transparent sheet, has the advantages of simple structure and convenient use, avoids the use of additional devices such as a Mach-Zehnder (MZI) interferometer, and reduces the system cost and complexity.

The invention provides a semantic segmentation method, device and equipment and a computer readable storage medium. The method comprises the steps: obtaining a display operation for a three-dimensional model; in response to the display operation, displaying the three-dimensional model on a human-computer interaction interface, wherein the human-computer interaction interface comprises semantic segmentation options; obtaining a selection operation for the semantic segmentation options; and in response to the selection operation, after a two-dimensional segmentation result of a two-dimensional image is obtained, displaying a semantic segmentation result of the three-dimensional model on the human-computer interaction interface, wherein the semantic segmentation result is determined accordingto the two-dimensional segmentation result of the two-dimensional image and the model attribute of the three-dimensional model, and the two-dimensional image and the three-dimensional model belong tothe same scene. Through the semantic segmentation method based on artificial intelligence provided by the invention, the segmentation efficiency of the three-dimensional model can be improved, and the user experience is improved.

The invention discloses an image feature representation and human body motion tracking method based on the second-generation strip wave transformation, which mainly solves the problem that the existing feature extraction is based on the human body contour and edge ambiguity, and cannot reflect the geometric flow direction and texture of the internal image. question. The implementation process is: input the video image to be processed, extract the frame diagram of human body parts; perform two-dimensional multi-scale wavelet transformation on the image; use quadtree division and bottom-up fusion rule to find the optimal geometric flow direction; quantify the optimal The geometric flow direction signal is transformed by one-dimensional wavelet and reorganized into two-dimensional form to obtain the Bandelet2 coefficient matrix; the maximum geometric flow statistical feature is extracted as the final image feature representation; the mapping relationship between image features and three-dimensional motion data is learned through the regression process, and the new Predict and recover 3D pose from training video images. The invention has the advantages of fast tracking, accurate result, small amount of calculation, reduced blurring of image representation, and can be used for human target recognition and three-dimensional posture reconstruction.

The invention relates to an attitude determination system and method of a small unmanned aerial vehicle, which have the advantages of low cost, low consumption, simple algorithm and small calculated amount. The system comprises an angular rate gyro chip, a triaxial electronic compass chip, a triaxial accelerometer chip and a microcontroller chip, wherein the angular rate gyro chip is used for measuring rolling, pitching and yawing angular rates of the unmanned aerial vehicle; the triaxial electronic compass chip is used for measuring a flight path azimuth angle, namely a yawing angle, of the unmanned aerial vehicle in a flight process; the triaxial accelerometer chip is used for measuring a component of the gravity in a body coordinate axis of the unmanned aerial vehicle; the microcontroller chip is connected with the angular rate gyro chip, the triaxial electronic compass chip and the triaxial accelerometer chip through I<2>C buses and is used for acquiring measuring data of sensors and determining the real-time attitude of the unmanned aerial vehicle according to the acquired data and the attitude determination method. The method comprises the steps of establishing an initial direction cosine matrix between a body coordinate system and an inertial coordinate system of the unmanned aerial vehicle, updating an attitude direction cosine matrix, standardizing the direction cosine matrixes and counting a corrected triaxial angular speed and triaxial attitude angles of the unmanned aerial vehicle.

The invention provides an electronic gear shifting device and relates to electronic gear shifting technologies. The electronic gear shifting device comprises a Hall sensor unit, a gearbox control unit, an electronic gear shifting control unit and a man-machine interaction unit, wherein the Hall sensor unit is used for detecting operation of a driver to a gear shifting handball; the gearbox control unit is sued for conducting the gear shafting action and detecting the gear of a gearbox of a vehicle; the electronic gear shifting control unit is used for acquiring the gear shifting intention of the driver through the Hall sensor unit and judging whether the gear shifting intention of the driver is dangerous, and/or the danger level of the gear shifting intention of the driver on the basis of the obtained gear of the gearbox and the obtained speed of the vehicle, and the electronic gear shifting control unit controls the gearbox control unit to carry out corresponding gear shifting operation to respond the judgment that the gear shifting intention of the driver is safe, and the judgment is made by the electronic gear shifting control unit; and the man-machine interaction unit can give an alarm to the driver, and the alarm is used for respond the judgment that the gear shifting intention of the driver is dangerous and/or the danger level, and the judgment is made by the electronic gear shifting control unit. When it is judged that the operation of the driver is dangerous, alarming or prompting is conducted through the man-machine interaction unit.

The invention discloses a small current system grounding fault positioning method. The method includes the following steps: extracting the zero sequence currents prior to and after the moment of a fault of each feeder terminal on a faulty feeder; moving a data window point by point and calculating the feature value of zero sequence current of respective feeder terminal under each data window; establishing a feature value curve for each feeder terminal; acquiring the sampling point corresponding to the maximum feature value of each feature value curve of the feeder terminal which is closest tothe bus on the faulty feeder; acquiring the feature values that correspond to the same sampling points that are acquired from the feature value curves of the rest of the feeder terminals; calculatingthe Euclidean distance of the feature values of the adjacent two feeder terminals; ranking the Euclidean distances from long to short and selecting N biggest Euclidean distances; if one Euclidean distance is greater than the sum of the rest of the N-1 Euclidean distances, taking the interval between two feeder terminals that correspond to the Euclidean distance as a fault interval. According to the invention, the method herein can increase reliability of fault positioning and reduce communication pressure of data channels.

The invention provides a data aggregation rendering method, a device, a system, electronic equipment and a storage medium. The method is applied to a server. The method comprises the steps: receivingdata rendering request information transmitted by a client; wherein the request information comprises multiple pieces of map fragment information corresponding to a current visual range; aggregating the plurality of data points according to the information of the plurality of map fragments and the coordinate information of the plurality of data points to obtain a first class center point set corresponding to each map fragment; wherein each first type center point is formed by aggregating at least one data point. Compared with a mode of transmitting all data points to the client, the method issimple; according to the method, the aggregated central point is transmitted to the client after data aggregation, so that the data volume transmitted to the client by the server and the data processing amount of the server are reduced, the bandwidth occupancy rate and the CPU resource consumption rate can be reduced in the whole process, and the rendering efficiency and the data display effect ofthe data points are improved.