41results about How to "Real-time performance" patented technology

An omnidirectional sensor array system, for example a panoptic camera, comprising a plurality of sensors arranged on a support of predetermined shape to acquire data, wherein said sensors are directional and wherein each sensor is attached to a processing node which comprises integrated electronics that carries out at least a portion of the signal processing algorithms locally in order to reduce the computational load of a central hardware unit.

A system and method combines videos for display in real-time. A set of narrow-angle videos and a wide-angle video are acquired of the scene, in which a field of view in the wide-angle video substantially overlaps the fields of view in the narrow-angle videos. Homographies are determined among the narrow-angle videos using the wide-angle video. Temporally corresponding selected images of the narrow-angle videos are transformed and combined into a transformed image. Geometry of an output video is determined according to the transformed image and geometry of a display screen of an output device. The homographies and the geometry of the display screen are stored in a graphic processor unit, and subsequent images in the set of narrow-angle videos are transformed and combined by the graphic processor unit to produce an output video in real-time.

A network switch, configured for performing layer 2 and layer 3 switching in an Ethernet (IEEE 802.3) network without blocking of incoming data packets, includes network switch ports, each including a flow module configured for generating a packet signature based on layer 3 information within a received data packet. The flow module generates first and second hash keys according to a prescribed hashing function upon obtaining first and second portions of layer 3 information. The flow module combines the first and second hash keys to form the packet signature, and searches an on-chip signature table that indexes addresses of layer 3 switching entries by entry signatures, where the entry signatures are generated using the same prescribed hashing function on the first and second layer 3 portions of the layer 3 switching entries.

A method for tracking coronary artery motion includes constructing (11) a centerline model of a vascular structure in a base phase image in a sequence of 2D images of coronary arteries acquired over a cardiac phase, computing (12), for each pixel in a region-of-interest in each subsequent image, a velocity vector that represent a change in position between the subsequent image and base phase image, calculating (13) positions of control points in each phase using the velocity vectors, and applying (14) PCA to a P×2N data matrix X^T constructed from position vectors (x, y) of N centerline control points for P phases to identify d eigenvectors corresponding to the largest eigenvalues of XX^T to obtain a d-dimensional linear motion model {circumflex over (α)}_p, in which a centerline model for a new image at phase p+1 is estimated by adding {circumflex over (α)}_p to each centerline control point of a previous frame at phase p.

A system is presented that provides real-time performance for iterative multi-user detectors, such as Turbo MUDs, which are used to separate simultaneous transmissions on the same frequency, by permitting the MUD to use a less computationally intense, fast-processing algorithm and to correct for errors caused by the fast processing. In order to reduce the errors, a voting system is coupled to the output of the multi-user detector within the iterative system. The voting system provides confidence values on a bit-by-bit basis for the estimates made by the multi-user detector, with the confidence values then being utilized as soft inputs to a bank of conventional single-user decoders.

The invention discloses a human body action recognition method and a mobile intelligent terminal. The human body action recognition method comprises the steps that human body action data are acquired for training so that feature extraction parameters and template data sequences are obtained, and the data requiring performance of human body action recognition are acquired in one time of human body action recognition so that original data sequences are obtained; feature extraction is performed on the original data sequences by utilizing the feature extraction parameters, and the data dimension of the original data sequences is reduced so that test data sequences after dimension reduction are obtained; and the test data sequences and the template data sequences are matched, and generation of human body actions corresponding to the template data sequences to which the test data sequences are correlated is confirmed when the successfully matched test data sequences exist. Dimension reduction is performed on the test data sequences so that the requirements for the human body action attitudes are reduced, and noise is removed. Then the data after dimension reduction are matched with the templates so that calculation complexity is reduced, accurate human body action recognition is realized and user experience is enhanced.

A method and apparatus for compressing and decompressing an audio signal. The apparatus comprises an input for receiving an audio signal derived from a spoken utterance, the audio signal being contained into a plurality of successive data frames. A data frame holding a certain portion the audio signal is processed to generate a feature vector including a plurality of discrete elements characterizing at least in part the portion of the audio signal encompassed by the frame, the elements being organized in a certain sequence. The apparatus makes use of a compressor unit having a grouping processor for grouping elements of the feature vector into a plurality of sub-vectors on the basis of a certain grouping scheme, at least one of the sub-vectors including a plurality of elements from the feature vector, the plurality of elements being out of sequence relative to the certain sequence. The plurality of sub-vectors are then quantized by applying a vector quantization method.

An abnormal action detecting device and method for detecting an abnormal action from a moving picture. An abnormal action detecting device (11) creates from-to-frame difference data from moving picture data inputted from a video camera (10), extracts feature data from three-dimensional data composed of frame-to-frame difference data by using a stereoscopic high-order local cross-correlation, computes the distance between a partial space based on a main component vector determined by a main component analysis technique from the past feature data and the latest feature data, and judges that an action is abnormal if the distance is greater than a predetermined value. By learning a normal action as a partial space and detecting an abnormal action as a deviation from the normal one, for example, even if several persons are present in the screen, an abnormal action of a person can be detected. The computational complexity is low and the real-time processing is possible.

The present invention discloses an unmanned aerial vehicle aerial image defogging algorithm based on a dark channel. The algorithm comprises the steps of: obtaining a dark channel image according to acolor foggy image, obtaining a global atmospheric light composition A according to the dark channel image, obtaining a coarse-transmissivity graph according to the dark channel image, performing downsampling of the coarse-transmissivity graph, performing guidance filtering, performing upsampling to obtain a fine-transmissivity graph, obtaining an initial defogging image according to the fine-transmissivity graph and the global atmospheric light composition A, and employing restriction contrast adaptive histogram equalization to perform local contrast enhancement of the initial defogging imageto solve the problems that the image is dark and highlights the local information of the original image after defogging of the dark channel and to be used for later image processing work. The unmanned aerial vehicle aerial image defogging algorithm is especially suitable for defogging operation of high-resolution foggy images shot by the unmanned aerial vehicle, can obtain more image detail points of the unmanned aerial vehicle while rapidly removing the fog images of the images of the unmanned aerial vehicle, can simplify computing and has good robustness.

The invention discloses a method for synthesizing a continuous and smooth panoramic video in real time by using discrete cubic panoramas. The method comprises the following steps of: carrying out feature point detection and feature matching on two input continuous cubic panoramas to establish triangular mesh dissection of the cubic panoramas on a spherical domain so as to realize preprocessing of the cubic panoramas; and carrying out real-time interpolation on the cubic panoramas, estimating the composition of a triangular mesh to be interpolated in a new camera position on the basis of the existing triangular mesh correspondence, and carrying out color filling on the new perspective triangular mesh on the basis of a homography matrix. Compared with the prior art, the method disclosed by the invention has the advantages of projecting the cubic panoramas into the spherical domain, adopting a matching-triangulation-interpolation strategy in the spherical domain, simplifying the complexity of interpolation processing and effectively improving the visual effect of interpolation.

A vision based hand posture recognition method and system thereof are disclosed. The method comprises the following steps of receiving an image frame; extracting a contoured hand image from said image frame; calculating a gravity center of said contoured hand image; obtaining contour points on a contour of said contoured hand image; calculating distances between said gravity center and said multiple contour points; recognizing a hand posture according to a first characteristic function of said multiple contour points. In embodiment, the finger number and hand direction of the hand posture can be determined according to the number and location of at least one peak of the first characteristic function.

A system and method for processing a plurality of channels, for example audio channels, in parallel is provided. For example, a plurality of telephony channels are processed in order to detect and respond to call progress tones. The channels may be processed according to a common transform algorithm. Advantageously, a massively parallel architecture is employed, in which operations on many channels are synchronized, to achieve a high efficiency parallel processing environment. The parallel processor may be situated on a data bus, separate from a main general purpose processor, or integrated with the processor in a common board or integrated device. All, or a portion of a speech processing algorithm may also be performed in a massively parallel manner.

The invention discloses a dangerous and old building monitoring system based on the Beidou positioning technology. The dangerous and old building monitoring system comprises a, an inclination and settlement monitoring module which is used for extracting the inclination characteristic quantity of the monitored house building in real time; b, a crack monitoring module which is used for monitoring the crack width of the monitored house building in real time; c, a stress-strain monitoring module which is used for performing real-time stress-strain characteristic quantity monitoring on the monitored house building; d, a dynamic characteristic monitoring module which is used for monitoring the change of the dynamic characteristic after internal damage of the monitored house building; e, an information processing module which is used for receiving the monitoring data outputted by the inclination and settlement monitoring module, the crack monitoring module, the stress-strain monitoring moduleand the dynamic characteristic monitoring module; f, a BDS positioning system; g, an auxiliary monitoring module; h, a central processing module; i, an intelligent analysis cloud platform; and j, a client side module. The advantages are that safety warning of the dangerous and old building can be realized.

A resolution of a low resolution depth image is increased by applying joint geodesic upsampling to a high resolution image to obtain a geodesic distance map. Depths in the low resolution depth image are interpolated using the geodesic distance map to obtain a high resolution depth image. The high resolution image can be a gray scale or color image, or a binary boundary map. The low resolution depth image can be acquired by any type of depth sensor.

The invention provides an abnormal behavior real-time monitoring method based on deep learning. The abnormal behavior real-time monitoring method comprises the steps of obtaining human body target video data through a camera; extracting the human body target video data frame by frame according to the interception frequency to obtain a human body target image, and inputting the human body target image into a core controller; enabling a core controller to extract body key points and a skeleton structure of the human body target image, perform cascade connection on the extracted body key points of each target according to a motion time sequence to obtain a body posture evolution graph, and input the body posture evolution graph into a classifier; wherein the classifier is based on a body posture evolution diagram; distinguishing action classification to acquire target actions. An artificial intelligence algorithm is applied to a traditional video monitoring system, improvement of functions and performance is obtained, classification information of actions can be rapidly generated, the state of a human body can be well fed back, and the abnormal behaviors of the human body are corrected and supervised.

The invention provides a peer-to-peer network management system and method. The method comprises the following steps: a management station interacts with finite peer nodes in a peer-to-peer network through agent to obtain the related information from the peer-to-peer network or peer nodes adopting the resource locate and discovery (RELOAD) protocol; the peer-to-peer network or peer nodes are set or a management command is given to the peer-to-peer network or peer nodes; and the peer-to-peer network or peer nodes can instantly report the related information such as performance or alarm, and the performance requirement of the system is ensured. Therefore, the effective monitoring, diagnosis, debugging and other managements of the peer-to-peer network can be realized, the management efficiency of the peer-to-peer network can be increased, the working performance of the management station can also be increased and the timely management of the peer-to-peer network can be finally ensured, thus the normal operation of the peer-to-peer network can be ensured.

The invention provides a data recorder based on HDLC (high-level data link control) and a recording method. The data recorder comprises a central processing module, a data storage module and an HDLC interface module, wherein the data storage module is used for storing data to be stored by the data recorder; the central processing module is used for receiving the data from an HDLC bus through the HDLC interface module, processing the data and sending the processed data to the data storage module; the HDLC interface module is used for receiving the data from the HDLC bus and sending the data to the central processing module. According to the data recorder provided by the invention, the problems in the prior art are solved; running data and fault data of a locomotive using the HDLC bus can be recorded in real time, so that a worker can perform immediate and quick analysis and remove faults.

A sensorized garment is provided. The sensorized garment includes an wearable element on a skin having a hand-tightening or a closure device to securely constrain sensors, for example electrodes to detect electrophysiological signals, during a physical activity. The wearable element has a three-dimensional shape to uniquely define a right side and a left part of a circumferential zone when the wearable element is worn. The sensorized garment further includes a band carried by the circumferential zone, and at least one conductor connected between the sensor and a zone of the sensorized garment.

The invention relates to a VR (Virtual Reality) visualized platform, comprising a VR view editing module, a VR view releasing module, a construction scheme editing module, a construction scheme management module and a site management module. The VR visualized platform of the invention has the advantages that: unified and centralized management is performed at one platform through the site management module, visualized inspection construction schemes and VR views are integrated into the unified platform, so as to achieve organic combination between construction schemes of transformer substations and the VR; in addition, a real scene can be simulated according to a VR panoramic view, technicians can lead project work managers and operators to get familiar with an on-site environment and tease out site work dangerous points and safety precautions; and moreover, the significance of civilized construction warning signs can be explained, technical measures for ensuring safety can be emphasized, safe working awareness can be improved, potential accident risks can be avoided, safety measures of various on-site working faces and various profession intersected faces can be defined, comprehensive safety technology disclosure can be performed, and an early-stage preparation efficiency of a project can be improved.

The invention provides a paint film automatic regulation and control system based on cascade regulation. The paint film automatic regulation and control system based on cascade regulation comprises avehicle speed regulation module, a paint film outer diameter PID module, a furnace temperature PID module and a fan PID module, wherein the paint film outer diameter PID module, the furnace temperature PID module and the fan PID module are connected with the vehicle speed regulation module in a cascade manner; the vehicle speed regulation module outputs the vehicle speed information to the paint film outer diameter PID module, the furnace temperature PID module and the fan PID module; the paint film outer diameter PID module controls the outer diameter of a paint film of a product according tothe vehicle speed information; the furnace temperature PID module controls the furnace temperature according to the vehicle speed information; and the fan PID module controls the concentration of thesolvent in the furnace according to the vehicle speed information. The paint film automatic regulation and control system can be used for timely and finely controlling the performance of the enameledwire product.