398results about How to "High real-time requirements" patented technology

The invention discloses a method for controlling game characters based on a human face analyzing technology. The method comprises the following steps: preprocessing an image obtained from an image input device simply and effectively; detecting the human face and positioning key points by a statistical learning method; and analyzing the positioning result to obtain human face gesture and expression information, and mapping the information to corresponding game control commands to achieve the real-time control function of the faces of the game characters. Using human face information of a player to control the game characters extends the interaction modes of the prior games by only shooting a picture of the human face of the player by a camera, analyzing the state and the expression information of the picture in a space, and converting the analyzed result into the control commands of the game characters. A video detection method of the invention is real-time, robust and easy to realize and operate. The method can enable game players to perform interactive operations in a more natural and more intellectual new mode, such as adopting self head gestures, face expressions and the like to endow games with more interactions and immersion.

The present invention relates to a robot hybrid system application framework based on a multi-core processor architecture. In the robot system with ARM / X86 multi-core processor as the controller, multi-core parallel processing architecture of the ARM / X86 multi-core processor is used to run the robotic hybrid system application framework comprising real-time operating system, non-real-time operating system and system supporting frame in the whole robot controller, so as to provide improved operating system services. In this application framework, a real-time operating system runs independently in one ARM / X86 core, while several non-real-time operating systems run on other ARM / X86 cores. The operating systems occupy processor resources and peripherals separately and run robotic applications with different real-time requirements. The application program can be used as a unified robot operating system (ROS) application node.

Provided is a parallel multi-core FPGA digital image real-time zooming processing method and device. The method comprises the steps that original video image data to be processed are acquired firstly. The original video image data are inputted to an FPGA chip and the number of zooming cores is confirmed, and the original video image data are divided into image sub-blocks having the same number with that of the zooming cores. Then the divided data are stored in a buffer, the data in the buffer are returned to the FPGA and parallel zooming processing is performed on each image sub-block via one zooming core respectively so that zooming is completed. All the image sub-blocks after zooming are displayed after splicing processing. The method has certain guiding effect on high throughput and high real-time application of digital video images.

The invention discloses a streaming media distribution and transmission network system, which comprises a network television center business platform, one or more node data centers and one or more user equipment, wherein one data center is used for receiving element information and streaming media data from the network television center service platform and sending the element information to the node data center; the address of the corresponding node data center is returned to the user equipment according to a domain system DNS (Domain Name Server) domain analysis request of the user equipment; the node data center is used for acquiring the streaming media data from the data center according to the element information; the streaming media data or the addresses of the other user equipment are returned when a content access request of the user equipment is received; the user equipment is used for sending the content access request to the corresponding node data center according to the address of the node data center returned by the data center; and the node data center interacts the data with the corresponding user equipment if the node data center returns the addresses of the other user equipment.

The invention discloses a hierarchical target allocation method for multiple unmanned aerial vehicle formations. The method is characterized by comprising the following steps of 1, clustering a plurality of ground targets to form a plurality of target clusters according to the number of the unmanned aerial vehicle formations; 2, regulating the obtained target clusters to keep the number of the ground targets in each target cluster consistent; 3, allocating the target clusters to the unmanned aerial vehicle formations in a one-to-one correspondence way by using an inter-unmanned aerial vehicle formation target allocation model and an inter-unmanned aerial vehicle formation target allocation algorithm; 4, allocating the ground targets to each unmanned aerial vehicle in the unmanned aerial vehicle formations by using an intra-unmanned aerial vehicle formation target allocation model and an intra-unmanned aerial vehicle formation target allocation algorithm. According to the method, the target allocation efficiency can be improved, and the problem of excessively long calculation time during large-scale target allocation for the multiple unmanned aerial vehicle formations can be effectively solved, so that the requirements of an application scenario such as a battlefield with a higher requirement on real-time performance are met.

The invention discloses a face detection method based on a convolutional neural network. The face detection method comprises a training stage and a testing stage, wherein the training stage comprises the following steps: inputting a training sample into the convolutional neural network, and obtaining a connection weight and an offset value of the convolutional neural network. The testing stage comprises the following steps: reading a video image; when a moving object is detected, extracting an interested area, moving for one pixel each time by utilizing a block with a n*n pixel size, and carrying out partitioning processing to the interested area to obtain a plurality of pictures; zooming the partitioned pictures to a size which is the same with the size of the training sample; and inputting the pictures into the convolutional neural network which finishes training for classification, wherein n is greater than or equal to 50 and less than or equal to 70. The convolutional neural network is used as a classifier, so that detection precision and speed can be improved.

The invention relates to a robot remote control welding system and a method used for remote welding, which relate to a robot remote control welding system and a method. The invention has the goal of solving the problems that in the existing remote control welding process, the real-time requirement of the welding process is high, the adaptability of the autonomous control remote control welding is low, the visual feedback of the remote welding is poor, and the system universality on welding tasks is low. The system has the technical scheme that a center monitoring human-machine interface is connected with a robot controller, and the robot controller is respectively connected with a laser visual controller, a robot and a welding power supply. The method has the technical scheme that a binocular camera transmits a welding environment image to a three-dimensional visual display, operators obtain the deep information of the remote end environment and guide a welding gun to move to the welding seam starting position, and four work modes: the direct control work mode, the remote control teaching work mode, the laser visual display tracking work mode or the human-machine cooperation sharing control work mode can be respectively adopted for completing the remote control welding. The invention is used for robot remote control welding.

The invention relates to a networking topologic structure of fleet formation and a design method for a combined multi-address system, belonging to the technical field of the aviation data link, radio navigation and aircraft automation. The invention provides a multi-address system and a networking topologic structure for networking communication and measurement of fleet formation, meets the passive detection and location and the cooperative attack task of the fleet formation to the moving-target. The invention provides the networking topologic structure which can realize the combined multi-address system and support the fleet formation on a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array) apparatus and a radio frequency channel on an intermediate-frequency signal processing circuit board of a fleet link terminal. The invention provides two novel multi-address access systems of communication topologic configuration, namely the CDMA Code Division Multiple Access) / FDAM combined multi-address communication system of the fleet formation based on cross-link networking topology and a graded hierarchical multi-address communication system of the fleet formation based on tree topology. The former can support a single-hop access mode of total network coverage and the latter provides a multiple-hop access mode of total network coverage.

The invention discloses a thrust distribution method for a power positioning system of an offshore drilling platform, which solves the problem of thrust distribution optimization of the power positioning system by applying a particle swarm algorithm through taking minimized propulsion system power consumption and a thrust error as target functions of the thrust distribution optimization and considering the thrust, thrust change rate, azimuth change rate and constriction conditions of a thrust forbidden sector. The method comprises the following steps of setting input quantity of the thrust distribution, including vertical resultant force, transverse resultant force and yawing moment, as given parameters; setting the thrust amplitude and direction of each thruster as unknown variables, namely, solutions to be optimized; analyzing platform stress, wherein the resultant forces and resultant force moments generated by all the thruster are equal to the input quantity of the thrust distribution; and building a thrust distribution mathematical model according to the distribution of offshore drilling platform thrusters and applying the particle swarm algorithm to solve the problem of thrust distribution optimization. The thrust distribution method has no special requirements to the target functions of the thrust distribution and less parameter to be regulated, is simple in operation, easy for realizing and fast in calculation speed, so that the requirement of high real-time capability of the power positioning system is fulfilled.

The control system includes VME bus chassis, CPU main board, Ethernet comm interface board (CIB), reflection memory CIB, PROFIBUS CIB, digital signal interface board, analog signal interface board, and electrical motor control board (MCB). Boards and cards are inserted to backboard inside the VME bus chassis. Through VME bus, boards and cards in backboards inside same chassis can communicate each other through VME bus. Each board contains VME bus controller. VME bus controller implements functions of gating board, bus arbitration, interrupt response control, and reading and writing data. Each board possesses own address. Through VME bus, gated board exchanges data to CPU board. Using shared memory exchanges data between CPU boards in same backboard bus. The reflection memory CIB executes remote comm and real time control among multiple control systems. MCB can exchange data to CPU main board, and control motor through optical fiber interface driven by power switch in bottom layer.

This invention discloses a micro controller with bus and intercommunicated nodes for multi-host communication system, which contains data transmission unit, receiving unit, asynchronous series communication interface, bus monitor check unit, priority mark transmission unit, bus priority arbitration unit, priority mark filter unit, data / address identifying unit, wherein the asynchronous series communication interface connected with a multi-host interface whose another end connected with said bus. Said invention not only realizes point-to-point multi-host communication among plurality of equipment but also does not increase the occupation to software and hardware resource of MCU, and reduces cost and raises efficiency.

A secure routing method of the converged network of a security-increased cellular network and a self-organized network is a converged network scheme and a routing secure method of the security-increased cellular network and an Ad hoc (self-organized network). The converged network comprises three main parts: a dual mode mobile terminal, a base station and a security agent, and adopts the converged network secure routing method to carry out routing; wherein, the dual mode mobile terminal comprises a smart phone, PDA, PC and a notebook computer; the base station stores the local terrain in detail to a maximum extent and locates the lawful nodes in the coverage area precisely to form network topology information; the security agent is connected with the nodes in the base station managing district by a main trunk path of the cellular network; the security agent is internally provided with a reliable center and an authentication center, which is used for managing the security of the converged network; when nodes give off routing requirements, the routing selection is executed in a 'first choosing an Ad hoc mode, second relying on a base station' way, and the routing is executed in the secure routing method of the converged network.

The invention provides an onboard distributed type POS real-time transmission alignment method based on federal filtering. An onboard distributed type POS is composed of a high-precision main POS and a plurality of secondary IMUs. Aiming at the problem that the instantaneity requirements of onboard distributed type POS real-time transmission alignment are high, the federal filtering based on a dispersion structure is used for transmitting and aligning. Firstly, a federal filterer structure design is carried out, namely two secondary filters and one main filter are designed; secondly, mathematic models of the two secondary filters are established to carry out Kalman filtering to obtain partial estimation; then, the main filter is used for estimating an overall situation filtering solution, and resetting the two secondary filters and the main filter; and finally, the positions, speeds and postures of the secondary IMUs are more accurately calculated by using a reset estimation state amount.

The invention relates to an automatic driving automobile track dynamic planning and tracking method based on transverse and longitudinal coordination, the method comprises the following steps of firstly, intensively sampling a motion state in a Frenet space in which an automobile runs to finish transverse and longitudinal local track replanning; in transverse and longitudinal trajectory coordination planning, considering that the automobile in a complex traffic environment and static and dynamic obstacles in the environment cannot be simply equivalent to mass points; introducing a hyperplane theory to evaluate whether a vehicle track interferes with static and dynamic obstacles in a surrounding traffic environment or not and evaluate the risk of collision, and designing the track evaluation index systems such as smoothness, comfort and driving efficiency of the track to evaluate the track quality. In the trajectory tracking stage, a Brush tire model is adopted to design an automatic driving vehicle dynamics manipulation limit constraint rule, and the model predictive control (MPC) is utilized to complete automatic driving vehicle active steering (AFS) and direct yawing moment control (DYC) integrated trajectory tracking control.

Provided are a network analytic system and method based on the mass data real-time processing technology. The system comprises a plurality of data adaptive nodes and data analytic clusters which are composed of a plurality of data analytic nodes and arranged in a network in a distribution mode, wherein the data analytic nodes support a P2P networking mode and a load balancing mechanism, so that the data analytic clusters have a telescopic function; the flow line type analytic processing process is completed among the data analytic nodes through an incident mechanism. According to the network analytic system and method supporting the real-time mass data processing, mass network data including operations of network fault monitoring, statistics, trouble shooting, diagnosis and the like can be analyzed and processed in real time, the network data can be analyzed in a fine mode, distribution type dynamic expansion is supported to expand system functions, and users can expand demands of themselves according to the analytic types to be detected in a self-defined manner. Furthermore, a distribution type structure is utilized by the system without excessively depending on single hardware performance, and the system can complete processing of complex logics of network data analysis and the like better. The system and method support multiple types of processing logics and lower the requirements for professional qualification of developers.

The invention relates to a multi-link concurrent access multimedia data transmission method. The present method that using a single operator to provide transmission bandwidth can not support the high quality and real time transmission of multimedia service. The invention provides a multi-link concurrent access multimedia data transmission method which includes the server information acquisition, the segmentation of the data to be transmitted, the data transmission, and priority distinguished data transmission and data protection. Wherein, the data transmission includes an alternative test cycle and a data cycle; in the test cycle, the transmitting equipment test the performance of the data transmission link and the access link; in the data cycle, the transmitting equipment transmits data at each link according to the test result of the test cycle, thereby making full use of the link bandwidth resource and avoiding the link congestion. The multi-link concurrent access multimedia data transmission method of the invention adopts the transmission type by distinguishing the priority of different data at different priority; thereby ensure the transmission reliability of the high priority data.

The invention discloses a velocity look-ahead preprocessing method oriented to efficient processing of tiny line segments. The velocity look-ahead preprocessing method comprises the steps of dynamically and adaptively selecting the number of look-ahead preprocessing segments, and performing optimization preprocessing on the cutting velocity of each look-ahead segment to guarantee a high feed velocity and smooth join; the velocity look-ahead preprocessing method comprises the two steps of adaptive selection of the look-ahead segments and pre-calculation of the optimal join velocity; a numerical control system adaptively selects the look-ahead preprocessing segments by determining the positions of inflection points according to the processing path change trend; according to the path change trend of the look-ahead segments, firstly, the final velocity of each look-ahead segment is roughly estimated, and then the optimal join velocity of each look-ahead segment is accurately planned according to the instruction velocity, the rough estimated value of the final velocity, the accelerated velocity and the segment length constraint of each look-ahead segment. The velocity look-ahead preprocessing method oriented to the efficient processing of the tiny line segments is high in velocity, and capable of realizing efficient and smooth join of the feed velocity to meet the velocity control requirements of high-velocity digital numerical devices.

The invention discloses a data aggregation method based on compressed sensing in a wireless sensor network. The method includes: uniform clustering of the sensor network is performed, a node with the most residual energy is selected as a cluster head node, member nodes independently select whether to participate in sampling with the probability ptx, and the cluster head node always participate in sampling; then sampling nodes obtain original signals f and obtain sparse representations x thereof through transform of sparse transform bases, x are projected in a measuring matrix phi, sparse measuring signals y are obtained and sent to the cluster head node, and the cluster head node merges the collected measuring signals to a signal Y by employing vectorization operators and sends the signal Y to a fusion center; and finally the fusion center performs reconstruction on the signals one by one by employing an adaptive weight GPSR algorithm and recovers the sparse representations X thereof. According to the method, the characteristics of noise-containing signals, large data bulk, and high timeliness requirement of the wireless sensor network are completely achieved, the adaptive weight GPSR algorithm does not need to know the signal sparsity in advance, and all high-dimensional signals can be accurately reconstructed in a short period.

The invention discloses a frame rate up-conversion video coding method based on a deformation movement model and a system. The method comprises steps that whether the deformation movement model can be constructed is determined according to the information of four corners of a left upper corner, a right upper corner, a left lower corner and a right lower corner of a present block; frame rate up-conversion for the present block is carried out through employing a deformation movement model prediction mode, a bidirectional matching mode or a template matching mode according to a determination result, the movement information of the present block is acquired, movement compensation is carried out according to the acquired movement information, and the movement information comprises but not limited to a reference direction, a reference index and a movement vector. The method is advantaged in that a movement model of the present block is predicted through the deformation movement model in advance, the movement information of a smallest block is further calculated, the searching time of the movement vector is saved, the coding / decoding time and a code rate are greatly reduced, a coding / decoding speed and coding / decoding efficiency are improved, and the method can be widely applied to the video coding field.

The invention discloses a Kinect-based augmented reality three-dimensional registration method, which adopts a depth image restoration method fusing the RGB-D information to solve the calibration and the image alignment errors caused by the holes, jitter, edge instability and other problems in a depth image, and provides guarantee for the accuracy of the subsequent three-dimensional registration. A close-range mode automatic judgment method based on a depth histogram is adopted, and in a non-close-range mode, a three-dimensional registration method based on Fast ICP is adopted to calculate the pose of the camera; in the close-range mode, the pose of a camera is calculated by using a three-dimensional registration method of fusing Fast ICP (fast iterative closest point) and ORB (oriented fast and rotated brief), so that the problem of registration failure caused by the hardware limitation is solved, and an accurate and stable three-dimensional registration result can be obtained no matter in a non-close-range mode or a close-range mode. The method is low in algorithm complexity, is easy to understand and implement, is not affected by the illumination and the complex scenes, meets the real-time requirements of an augmented reality system, and can solve a mutual shielding problem.

The invention discloses a multiple-subarea matching method constraint by a space relation. A plurality of subareas are selected in a matching template as sub-templates, and the sub-templates are separately and simultaneously matched with to-be-matched images. Space position relations among the sub-templates are utilized to realize the accurate matching of the images. The method comprises the following steps: the plurality of subareas are selected in the matching template respectively as the sub-templates; the space relations among the sub-templates are determined; the plurality of sub-templates keep the space position relations unchanged to form combined templates; matching is carried out by utilizing the mobile searching performed by the combined templates on the to-be-matched images, and a plurality of similarity values of the combined templates are obtained; the plurality of similarity values are compared, and a position where a maximum similarity value exists is regarded as an optimum matching position to complete the matching. The method of the invention utilizes the mutual space relations among the plurality of sub-templates to reach the requirements for accuracy and precision of matching. The object identifying performance of the method is substantially improved in real-time performance compared with the conventional large-template gray scale or contour matching algorithms.

The invention discloses an on-orbit self-adaptive focusing method for a space optical camera, which comprises the following steps of: (1) using the space optical camera to shoot an image against the ground when the defocusing amount of the camera is adjusted once during an on-orbit automatic focusing process, and obtaining sequential images after shooting for multiple times; (2) calculating displacement amount between two adjacent images by a registration algorithm to find overlaps of the two adjacent images; (3) evaluating and calculating image definitions of the overlaps to obtain a plurality of groups of image definition evaluation values which takes two adjacent images as a group; and (4) mapping the definition evaluation values of a series of images to the same evaluation system by transmitting on a basis that every two adjacent groups of images have the same image definition evaluation value so as to find the best focusing position. The method can enable the space camera to finish automatic focusing within a short period of time during an on-orbit moving process, and is suitable for an application field with short focusing time and high focusing frequency.

The invention provides a hardware real-time simulation multi-channel multi-type communication protocol chip system and method and a medium, and the system comprises a system interaction module which is used for completing the connection with a tested system and monitoring the control instruction of the tested system in real time; A PL logic module in the FPGA which is used for finishing hardware acceleration of the protocol chip; An ARM1 module in the FPGA which is used for realizing a multi-protocol chip function; The FPGA internal ARM0 module which is used for realizing interaction with an upper computer and realizing multi-protocol chip control and data interaction; And the upper computer software platform module which is used for realizing function customization of a multi-protocol chip and real-time data monitoring and control of an analog chip. A user can carry out customized simulation according to different chip manuals, control over information such as a chip register is completed through C codes, customized updating is completed through a system software platform, and convenience and flexibility are achieved.

The invention discloses a method for realizing a KLT (Karhunen Loeve Transform) moving target tracking algorithm based on a multicore DSP (Digital Signal Processor). The method disclosed by the invention comprises the steps of: image smoothing, pyramid graph calculating, image gray value difference calculating, feature point selecting, feature point tracking and feature point consistency checking; and six modules for executing the steps are transplanted into the multicore DSP, wherein one core is allocated to each module, and the processing is carried out in a datastream manner. According to the method, a KLT tracking algorithm is adopted, and the advantage and processing speed of multicore parallel execution of the multicore DSP are used, so that the real-time and accurate tracking of moving targets can be realized.

A method of image information enhancement in radiography relates to image information processing techniques in radiography. The method comprising steps of: normalizing an acquired image A(x,y) to form a normalized image B(x,y); filtering the normalized image B(x,y) by a low-pass filter to obtain an filtered image C(x,y); calculating a relative standard deviation for each pixel in the image A(x,y), three times the relative standard deviation being an edge threshold for each pixel; thresholding a difference image obtained by subtracting the filtered image C(x,y) from the normalized image B(x,y) by using the edge threshold for each pixel to form a threshold-processed image D(x,y); enhancing a contrast of the threshold-processed image D(x,y) by using a non-linear function to form a contrast-enhanced image E(x,y); determining a enhancement coefficient a(x,y); obtaining a edge-enhanced image F(x,y) by multiplying the enhancement coefficient a(x,y), the contrast-enhanced image E(x,y) and the filtered image C(x,y); and generating a resulting image by multiplying a sum of the edge-enhanced image F(x,y) and the filtered image C(x,y) with the maximum value Amax As compared with the prior arts, the inventive method has a fast processing speed for image information enhancement and a simple algorithm, images clearly, eliminates noises in the images, and satisfies the requirements of relatively more enhancement to the contrast of the dark regions in the scanned images.

The invention discloses an FPGA-based enhanced serial port and working method thereof. The FPGA (Field Programmable Gate Array) is taken as hardware, and the serial port is constructed in the FPGA by using an HDL language; the Baud rate, check mode, data length and the like of the serial port are conveniently and quickly configured by an external program, and the serial port automatically receives and sends data; the size of a cache for transceiving the data only depends on the scale of the FPGA; and a plurality of serial ports can be generated by multiplexing through simple operation, and the processing time and hardware resource of a main program are not occupied. For the main program, it is similar to operation on an external storage, so the main program is freed from the frequently interruption or query operation, the running efficiency of the main program is improved, the stability of the system is improved and the transplantability and the expandability of software are improved.

The invention discloses a multichannel audio signal parallel collection device. The collection device comprises an FPGA processing unit, multiple audio decoding chips and a processor; the multiple audio decoding chips are electrically connected to the FPGA processing unit; the FPGA processing unit is electrically connected with the audio decoding chips and the processor and used for receiving audio data of all the audio decoding chips, namely audio data of multiple channels, fusing and recoding the audio data, increasing the transmission rate with the processor and transmitting the recoded data to the processor. Compared with a traditional integrated A / D chip scheme, cost and power consumption of a whole system are greatly lowered through the collection device.

The invention discloses a monocular multi-target identification and positioning method based on artificial marks. The monocular multi-target identification and positioning method based on artificial marks includes the following steps: according to set colors, performing binarization processing on an original image so as to obtain a binary image; extracting contours after performing preprocessing on the binary image; performing ellipse fitting on each contour, and rejecting the ellipse which cannot satisfy the shape requirement; combining every two ellipses, and taking the endpoints of transverses as the apexes to form quadrangles; calculating the evaluation coefficient (i)e( / i) of each quadrangle; analyzing the evaluation coefficients, and performing multi-target selection; and taking four apexes of each quadrangle as the feature points, using an iterative method to solve PnP, and successively obtaining the three dimensional pose, of a camera, relative to each mark. The monocular multi-target identification and positioning method based on artificial marks uses identification of transverses to replace identification of segments, and provides an evaluation coefficient aiming at multi targets, thus being able to effectively screening marked targets and solving the relative poses even in a complicated background, and having high robustness. Besides, the monocular multi-target identification and positioning method based on artificial marks has the advantages of being simple and practicable, having relatively lower time complexity in the algorithm, being able to obtain a calculation result with high frame rate, and satisfying the requirement of high timeliness.