66results about How to "Increased computational burden" patented technology

A method and system tracks objects using a surveillance database storing events acquired by a set of sensors and sequences of images acquired by a set of cameras. Sequences of temporally and spatially adjacent events sensed by the set of sensors are linked to form a set of tracklets and stored in the database. Each tracklet has endpoints being either a track-start, track-join, tracklet-merge or tracklet-end node. A subset of sensors is selected, and a subset of tracklets associated with the subset of sensors is identified. A single starting tracklet is selected. All sequences of tracklets temporally and spatially adjacent to the starting tracklet are aggregated to construct a tracklet graph. The track-join nodes and the track-split nodes are disambiguated and eliminated from the track graph to determine a track of the object in the environment.

The present invention relates to a device and method for monitoring for sleep disordered breathing or other types of disordered breathing such as Cheyne-Stokes breathing. More specifically, a device and method for detecting disordered breathing is provided that monitors a physiological parameter, which becomes cyclical due to apnea-hyperpnea (or arousal) alternation and provides the basis for the determination of a number of breathing disorder metrics.

In a virtualized computer system, a network frame is transmitted from a virtual machine using a network interface device, possibly through a virtual switch, by copying only a part of the network frame to the transmit buffers that have pre-translated mappings from guest physical addresses to hypervisor virtual addresses and to machine addresses. The length of the part of the network frame that is copied to the transmit buffers may be variable.

This invention deals with an advanced Real-time Grid Monitoring System (RTGMS) suitable for both single-phase and three-phase electric power systems. This invention provides an essential signal processing block to be used as a part of complex systems either focused on supervising and diagnosing power systems or devoted to control power processors interacting with the grid. This invention is based on a new algorithm very suitable for real-time characterization of the grid variables under distorted and unbalanced grid conditions. The main characteristic of this invention is the usage of a frequency-locked loop, based on detecting the grid frequency, for synchronizing to the grid variables. It results in a very robust system response in relation to existing technique based on the phase-angle detection since grid frequency is much more stable variable than the grid voltage / current phase-angle, mainly during grid faults. Moreover, the algorithm supporting this invention is very efficient and can be implemented in regular industrial microprocessors. These features make the RTGMS object of this invention ideal to be applied in the control of distributed generation systems (DGS), flexible AC transmission systems (FACTS), power quality conditioners (PQC) and uninterruptible power supplies (UPS). In all these systems, the fast and precise real time detection of the voltage and / or current sequence components under grid fault conditions is a crucial matter.

In an automatic fingerprint authentication or identification system, the fingerprint image acquisition is severely effected by the limitations of the acquisition process. The two modes of input, viz. scanning inked fingerprints from paper records or directly from a finger using live-scan fingerprint scanners suffer from the following noise sources in the input in addition to standard noise in the camera. Non-uniform ink application, uneven pressure while rolling on the paper or pressing on the scanner surface and external dirt like oil and climatic variations in the moisture content of skin are some of the main causes for the ridges and valleys not to be imaged clearly. This invention deals with a method of learning a set of partitioned least-sqaures filters that can be derived from a given set of images and ground truth pairs as an offline process. The learned filters are convolved with input fingerprint images to obtain the enhanced image.

A method and system tracks models and predict object movement in an environment. Sequences of temporally and spatially adjacent events sensed by the set of sensors are linked to form a set of tracklets. Each tracklet has an associated starting and terminating location. The tracklets are used to construct a directed graph including starting nodes, terminating nodes, and, intermediate nodes connected by edges. The intermediate nodes can be split nodes where tracklets diverge onto different tracks, and join nodes where multiple tracklets converge onto a single path. Probabilities are assigned to the edges to model and predict movement of the objects in the environment.

The invention belongs to the technical field of video target tracking, and provides an online video target tracking method based on depth cross similarity matching. The method comprises the followingsteps of designing a depth feature cross similarity module, capturing all local similarity information of a template and a sample, wherein obtained similar features are no longer sensitive to displacement and deformation; designing a similarity attention layer, distributing weight coefficients for cross similarity results of different spatial positions, and enabling a tracking algorithm not to respond to edge background interference; and designing a loss function containing a parameter regularization item, and rapidly optimizing the parameter to an optimal value. Based on the above three-pointbasic scheme, the deep learning twin network based on matching is used as a basic framework, and any target in the video is tracked in a precise and robust manner. From the tracking effect, the method provided by the invention has the capabilities of distinguishing similar objects, re-identifying a reappearing target after shielding and coping with rotation and deformation, and can be applied tovideo applications such as automatic driving of a front scene, autonomous flight of an unmanned aerial vehicle, traffic or a safety monitoring video and the like.

A computer system that implements a quality of service policy. Information defining the quality of service policy is stored in a central location, such as a server within an enterprise. Policy management software on clients in the network download quality of service policy information from the central location to the clients. Within a client, a portion of the policy information is selected based on its applicability to a particular connection. The selected information is cached with an association to the connection so that, as datagrams are generated for transmission over the network, relevant policy information is quickly accessed for use in controlling transmission characteristics of datagrams sent using that connection. Additionally, time information may be associated with information in the cache as well as policy information available from the server. Whenever a datagram is to be transmitted over a connection associated with out-of-date policy information in the cache, the cache information is updated.

An echo cancellation invention which does away with the need for an accurate double-talk detector, while maintaining a higher adaptation gain for quicker convergence and also providing increased stability. It does this by operating two filter models of the acoustic path in parallel. One model is adapted continuously to find the most accurate model of the echo path, while the other is not. The echo canceller output is taken from the filter that is not adapted. A comparison of the residual error (echo) is done between the model being adapted and the model being listened to at regular intervals. When the model being adapted has less error (echo) than the model being listened to its filter coefficients are copied to the other model. If the model being adapted has greater error (echo) than the other model (caused by noise diverging the adaptation process) then the adapted model has its coefficients overwritten by the filter that is listened to. This process keeps error from accumulating in the adapted filter by removing divergence at regular intervals (improving overall convergence). Since the output is taken from the filter that is not adapted, higher adaptation gains may be used without fear of the user hearing any effects of adapting with a higher gain in the presence of noise.

In a virtualized computer system, a network frame is transmitted from a virtual machine using a network interface device, possibly through a virtual switch, by copying only a part of the network frame to the transmit buffers that have pre-translated mappings from guest physical addresses to hypervisor virtual addresses and to machine addresses. The length of the part of the network frame that is copied to the transmit buffers may be variable.

The invention belongs to the information security technical field and provides an elliptical curve bypass attack-proof method based on the randomization tape symbol scalar multiplication. The elliptical curve bypass attack-proof method based on the randomization tape symbol scalar multiplication of the invention comprises the following steps that: firstly, a key in binary system is represented to be a randomization tape symbol binary number; secondly, a finite state machine for realizing the randomization tape symbol scalar product is obtained basing on a randomization addition and subtraction chains of the elliptical curve scalar product; finally, the elliptical curve scalar production operation is completed basing on the finite state machine and the improvement of the ability of the elliptical curve cryptograph to resist the bypass attack is realized. The invention which improves the security of the realization of the method can be applicable to the fields such as the cryptograph theory, the password cracking, the security chip design and the smart card design.

A method and system tracks models and predict object movement in an environment. Sequences of temporally and spatially adjacent events sensed by the set of sensors are linked to form a set of tracklets. Each tracklet has an associated starting and terminating location. The tracklets are used to construct a directed graph including starting nodes, terminating nodes, and, intermediate nodes connected by edges. The intermediate nodes can be split nodes where tracklets diverge onto different tracks, and join nodes where multiple tracklets converge onto a single path. Probabilities are assigned to the edges to model and predict movement of the objects in the environment.

This invention deals with an advanced Real-time Grid Monitoring System (RTGMS) suitable for both single-phase and three-phase electric power systems. This invention provides an essential signal processing block to be used as a part of complex systems either focused on supervising and diagnosing power systems or devoted to control power processors interacting with the grid. This invention is based on a new algorithm very suitable for real-time characterization of the grid variables under distorted and unbalanced grid conditions. The main characteristic of this invention is the usage of a frequency-locked loop, based on detecting the grid frequency, for synchronizing to the grid variables. It results in a very robust system response in relation to existing technique based on the phase-angle detection since grid frequency is much more stable variable than the grid voltage / current phase-angle, mainly during grid faults. Moreover, the algorithm supporting this invention is very efficient and can be implemented in regular industrial microprocessors. These features make the RTGMS object of this invention ideal to be applied in the control of distributed generation systems (DGS), flexible AC transmission systems (FACTS), power quality conditioners (PQC) and uninterruptible power supplies (UPS). In all these systems, the fast and precise real time detection of the voltage and / or current sequence components under grid fault conditions is a crucial matter.

A system and method for time-frequency analysis in which acquired signals are modeled as frictionally damped harmonic oscillators having a friction factor that is not a free parameter are provided. The friction factor is selected as a function of the frequency value of the associated oscillator, such that an increase in both temporal and spectral resolution are provided over existing time-frequency analysis methods. The friction factor is also selected to define a spectral band, within which the given oscillator can detect data oscillations. The properly selected friction factor thereby provides the analysis over a broad spectral range that can span many orders of magnitude.

A GPS-aided inertial navigation method includes providing multiple sensors including multiple inertial measurement units (IMUs) and at least one global positioning system receivers and antennas (GPSs) and computer with embedded navigation software. The computer interfaces with all IMUs and all GPS receivers; running, in parallel, in multiple standard inertial navigation (IN) schemes; computes the mean of all the INSs to obtain a fused IN solution for the IMUs' mean location; computes the mean of all the GPS solutions to obtain a fused GPS solution for the GPS antennas' mean location; applies a lever-arm correction, with the vector from the mean IMU location to the ‘mean antenna’ location, to the fused GPS solution; feeds the fused IN solution and the lever-arm corrected fused GPS solution to a single navigation filter, as if there were a single IMU and a single GPS; and runs an IMU / IN / GPS correction module.

A macroblock skip mode judgement method for an encoder applies to implementation of H.264 encoder hardware compressing a plurality of successive frames and comprises steps: undertaking IME (Integer Motion Estimation) for two frames appearing at neighboring time points to calculate MV (Motion Vector), and calculating PMV (Predicted Motion Vector) of the frame appearing the preceding time point; if MV equals PMV, modifying the cost function of MV to be a negative value; undertaking block encoding and compare the costs; if the cost of the Inter mode is smaller than the cost of the Intra mode, setting the mb (macroblock) skip flag to be 1; using Entropy to compress the mb skip flag, and omitting compressing the macroblock represented by the mb skip flag.

The invention discloses a voice emotion recognition system which comprises: an audio preprocessing module, a CNN module, a pyramid FSMN module, a time step attention module and an output module, whichare connected in sequence. The invention also discloses a speech emotion recognition method applied to the speech emotion recognition system, wherein the method comprises the following steps: 1, carrying out the preliminary operation of speech, thereby obtaining a speech spectrum feature map; 2, operating the spectrogram feature map, and constructing a spectrogram feature map containing audio shallow information; 3, further processing the spectrogram feature map containing the audio shallow information to obtain deeper semantic information and context information; 4, processing the spectrogram feature map with deeper semantic information and context information to obtain a feature vector with the highest relevancy between the whole speech and the emotion of the speaker; and 5, outputtingan emotion category corresponding to the whole section of voice. Compared with the prior art, the speech emotion recognition performance of the speech emotion recognition method is greatly improved.

The invention provides an LED display screen. Light-emitting tube arrays are uniformly arranged on the LED display screen in a column and row equidistant manner by taking red, blue, green and yellow light-emitting tubes or red, blue, green and indigo light-emitting tubes as display units, for each 2x2 light-emitting tube array, red, blue, green and yellow light-emitting tubes or red, blue, green and indigo light-emitting tubes form a square, and four light-emitting tubes in different colors are combined into four primary-color pixel points; a light-emitting tube is enabled to participate in the process of imaging of twelve adjacent pixel points through a simple virtual mapping relationship, so that display effect is smoother, calculation burden of a control circuit cannot be increased greatly, the number of the light-emitting tubes is reduced, and manufacturing cost is lowered.

The invention relates to a scene reduction-based intelligent power grid optimal scheduling method containing wind power and demand side resources, which comprises the following steps of: 1, predictingthe output power of a wind power plant on the next day, modeling a wind power prediction error, and randomly generating a plurality of initial scenes containing wind power output; step 2, performingscene reduction on a plurality of initial scenes containing wind power output randomly generated in the step 1 so as to generate a classic scene set; step 3, establishing an optimal scheduling model of the virtual power plant and the traditional unit considering uncertainty by taking minimization of the scheduling cost of the power system as a target; and 4, calculating the minimum value of the target function of the optimal scheduling model, and carrying out optimal configuration on the virtual electric field. According to the method, energy gathered by the virtual power plant can be effectively combined, fluctuation and uncontrollability of new energy generated output can be reduced during internal optimal scheduling, distributed power sources and loads are cooperatively managed, and safe and stable operation of a power grid is promoted.

An application contacts the ASIC with a request for a job, along with the name or identifier of a data stream to pattern match against, the name or identifier of the pattern set to use, and whether the job is partial or full. Depending on the priority rules set by the ASIC administrator, the ASIC may stop the job it is currently doing and begin work on the new job, or wait until the current job is finished before starting the new job. The ASIC determines if the pattern set for the new job is already stored in the cache, and contacts the calling application if it is not. Once the correct pattern set is loaded, the ASIC begins pattern matching on the requested data stream. The data stream is compared byte by byte with the each of the patterns in the loaded set. The ASIC will return a match to the calling application if a match has been made with one of the patterns in the pattern set.

Overlay error of a lithographic process is measured using a plurality of target structures, each target structure having a known overlay bias. A detection system captures a plurality of images (740) representing selected portions of radiation diffracted by the target structures under a plurality of different capture conditions (λ1, λ2). Pixel values of the captured images are combined (748) to obtain one or more synthesized images (750). A plurality of synthesized diffraction signals are extracted (744) from the synthesized image or images, and used to calculate a measurement of overlay. The computational burden is reduced compared with extracting diffraction signals from the captured images individually. The captured images may be dark-field images or pupil images, obtained using a scatterometer.

A fluorescence optical tomography system and method uses a photon migration model calculator for which absorption and reduced scattering coefficient values are determined for each source / detector pair. The coefficient values may be determined by measurement, in which a time resolved detector detects the excitation wavelength and generates temporal point spread functions from which the coefficient values are found. Alternatively, the coefficient values may be determined by calculating them from a dataset containing a spatial distribution of absorption and reduced scattering coefficients in a volume of interest. The fluorescence detection may be continuous wave, time resolved, or a combination of the two. An estimator uses a detected fluorescence signal and an estimated fluorescence signal to estimate the image values.

Disclosed is a PTZ (pan / tilt / zoom) camera focusing method and system. The method includes the steps of according to multiple focal length sampling points, multiple object distance sampling points and focal values of the focal length and object distance sampling points, acquiring a value b of a current focal length, b = y - Kx(f), with x(f) referring to the focal value x changing with change of the focal length f, y to an image distance, and K to a constant; according to the image distance y of the current focal length and the value b of the current focal length, calculating the focal value of the current focal length. The technical problem that the prior method of controlling the focal position of a camera by transmitting the focal value applies only to parfocal lens cameras and applies to no verifocal lens cameras is solved. In use of the PTZ camera focusing method and system in focusing, the cameras with two types of lenses both can accurately focus a tracked target to acquire clear target images.

The invention discloses a choosing method for a multi-point positioning location solution in the complex scene environment. The choosing method is mainly in charge of calculation of TDOA of a multi-point positioning system in the complex and actual scene environment to the target positioning location and the self-inspection output function of the positioning location. According to the main principle, follow-up processing and choosing operation are conducted on a positioning result by judging whether a chan's multi-point positioning algorithm two-step location estimation residual error sample conforms to probability distribution in the normal noise level or not; if yes, the current positioning result is directly output; if not, all the station points are removed sequentially, other m-1 station points are reserved, and positioning is conducted m times; a positioning combination with the maximum occurrence probability of the two-step location estimation residual error sample is selected; if the probability meets the preset threshold, a corresponding positioning solution is output; if the probability does not meet the preset threshold, the solution is removed. According to the choosing method for the multi-point positioning location solution in the complex scene environment, correlation of two location estimation values is taken into account, the calculation quantity is small, no other prior information is needed, universalization and simplification are facilitated, and the requirement of multi-point positioning engineering application in the complex scene environment can be met.

The invention relates to an LED display screen. Light-emitting tube arrays are formed on the LED display screen by taking red, blue, green and yellow light-emitting tubes or red, blue, green and indigo light-emitting tubes as display units, and space between columns is twice of space between rows, or vice versa; each 2x2 light-emitting tube array comprises red, blue, green and yellow light-emitting tubes or red, blue, green and indigo light-emitting tubes, four light-emitting tubes in different colors are combined into four primary-color pixel points, and a light-emitting tube is enabled to participate in the process of imaging of eight adjacent pixel points through virtual mapping,, so that resolution ratio of the display screen is guaranteed, color gamut of the display screen is increased, and manufacturing cost is lowered.

A method for motion estimation with respect to a sequence of images. Based on predictor vectors respective updates are generated. The updates are distributed over a plurality of tables. Only a single table of updates is applied to a given predictor vector to generate a limited set of candidate vectors only.

The invention discloses a control method for suppressing rectification start impact current of a three-phase LCL type grid connection converter. The control method also enables a system to have relatively good stability in the presence of a weak electric grid. The control method comprises the steps of detecting an AC side filtering capacitor voltage to serve as a synchronous voltage of a decoupled bisynchronous phase-locked loop; outputting a d-axis fundamental wave positive-sequence component of the filtering capacitor voltage by the decoupled bisynchronous phase-locked loop, and regarding the d-axis fundamental wave positive-sequence component and the filtering capacitor voltage as a feed-forward component; and adding the feed-forward component into a current loop of the grid connection converter. The start control performance f the LCL type grid connection converter is accordingly improved, and the system also can keep relatively strong robustness in the presence of the weak electric grid. On the basis of the traditional three-phase LCL type grid connection converter current control method, the control method for suppressing the rectification start impact current of the three-phase LCL type grid connection converter does not add the sampling amount, the hardware cost and the computational burden of the system.

An architecture for training the weights of artificial neural networks provides a global constrainer modifying the neuron weights in each iteration not only by the back-propagated error but also by a global constraint constraining these weights based on the value of all weights at that iteration. The ability to accommodate a global constraint is made practical by using a constrained gradient descent which approximates the error gradient deduced in the training as a plane, offsetting the increased complexity of the global constraint.

The invention relates to the technical field of artificial intelligence, and discloses a character recognition method and device based on a real plate, computer equipment and a storage medium, and themethod comprises the steps: starting a preset infrared light generator and a visible light generator; performing image acquisition processing on the real plate to obtain a first image; if the color value contrast of the second character pattern in the first image is greater than a preset contrast degree threshold, obtaining a second text; turning off the infrared light generator, and turning on apreset ultraviolet light generator; performing image acquisition processing on the real plate to obtain a second image; if the color value contrast of the first character pattern in the second imageis greater than a contrast degree threshold, obtaining a first text; and combining the first text and the second text to obtain a final character recognition result. Therefore, the accuracy of character recognition is improved. In addition, the invention also relates to a blockchain technology, and the character recognition model can be stored in the blockchain.