36results about How to "Obvious speed advantage" patented technology

The invention provides a deep learning image identification method and a deep learning image identification system used for intelligent driving, and a terminal device. The deep learning image identification system comprises a sharing convolutional network, an area segmentation network, and a target identification network. The area segmentation network is used for area classification processing based on a characteristic graph extracted by the sharing convolutional network, and the target identification network is used for target identification positioning processing based on the characteristic graph extracted by the sharing convolutional network. The sharing convolutional network is monitored by using the area segmentation result acquired by the area segmentation network and the target identification result acquired by the target identification network, and the sharing learning of the area segmentation network and the target identification network is completed. An obvious speed advantage on an aspect of multi-task learning is provided, and by comparing with the two independent networks learning individually, the deep learning image identification system has advantages of less consumed time and high efficiency, and in addition, a convolutional layer repetitive operation problem is effectively prevented, and multi-task detection and multi-task identification are completed.

A remote sensing image near-lossless compression hardware realization method based on improved JPEG-LS algorithm comprises the steps (1) directly calculating to obtain a pixel reconstruction value by the pixel actual value of an input image and accomplishing the calculation in a single clock period by using a formula Rx=int[Ix / (2Near+1)]*(2Near+1), in the formula, the Rx and the Ix are respectively the pixel reconstruction value and actual value, the int is rounding operation, and the Near is a compression ratio control factor; (2) calculating a context environmental variable Q according to the obtained pixel reconstruction value, if Q is equal to 0, performing run coding, otherwise, going to a step (3) of performing conventional coding; (3) calculating the predictive value of the current pixel by the pixel reconstruction value according to the geometric position relation between the current pixel and the adjacent pixel; (4) calculating the residual value between the predictive value and actual value of the current pixel; and (5) performing Golomb coding after quantising the obtained residual value, and synchronously updating the parameter variable corresponding to the context environmental variable Q using the quantisation result.

The invention discloses a GPU (graphics processing unit) processing method for a high-dimensional data stream in an irregular stream, and belongs to a processing method for the high-dimensional data stream. The method comprises the steps that a GPU-based data stream parallel computation model performs parallel computation processing on the high-dimensional data stream in the irregular stream of a data source and then sends the high-dimensional data stream to a client end; a CPU (central processing unit) host unit is a core for logic judgment and control and is used for serial computation; all or part of synopsis data structures are transferred into a video memory of GPU equipment; when the client end provides an inquiry request, the CPU host unit is in charge of calling a kernel program of the GPU equipment and executing an inquiry algorithm kernel on the synopsis data structures in the video memory of the GPU equipment, an inquiry result is returned to the CPU host unit, and finally the CPU host unit converts the result into a readable final inquiry result and returns the readable final inquiry result to the client end. According to the GPU processing method, the relevance between the high-dimensional data streams under a synchronous sliding window mode can be precisely identified on line; the outstanding speed advantage is realized, and the requirement on instantaneity of the high-dimensional data stream is well met.

The automatic state machine searching and matching method includes the following steps: establishing total match state list; inputting information to be matched; automatic searching and matching with the automatic state machine by means of the total match state list; and outputting the searching and matching result. The present invention has fast searching speed independent on regulation number and no special requirement on the code of searched key word, can search text as multilingual combination, and can search several specific key words simultaneously or some specified key word. The present invention may be used in real-time network information monitoring system, large scale monitoring system with great search information amount and common information detecting system.

The invention provides a single-machine minimized total tardiness scheduling method based on a distributed robust model and belongs to the technical field of production scheduling and internal and external production resource optimization. The method comprises the steps that a distributed robust optimization model for solving the single-machine scheduling problem is established firstly to obtain atarget function expression of the model; then, the distributed robust optimization model is converted into an integer second-order cone programming model; the converted model is solved, arrangement combinations of all workpiece machining sequences are expressed in a search tree in an enumeration way, the search tree is pruned through a branch-and-bound algorithm, and finally an optimal single-machine scheduling scheme of the minimized total tardiness is obtained. The method gives consideration to uncertain factors of a production environment, so that the model is a determined single-machine model more conforming to the actual production situation compared with an assumed production environment, and the obtained scheduling scheme can be better applied to actual production.

The invention belongs to the field of electric vehicles, and particularly relates to an integrated chassis control method of a four-wheel drive electric vehicle. The method comprises the following steps: utilizing a distributed driving electric vehicle kinetic model, namely a fourteen-degree-of-freedom vehicle control model, a UniTire tire model, a motor control model and a driver model, designinga vehicle state and parameter online estimation system, a side slip angle observer, a stability judgment system, an integrated coordination chassis control model, model prediction control and a multi-step center correction interior point optimization algorithm considering constraints; by analyzing the two-degree-of-freedom vehicle model, obtaining an off-line optimization system of a reference state trajectory and expected control variable. In the invention, two control methods of active front wheel steering and direct yawing moment control are combined, and the control stability of the vehicle is improved by optimizing the chassis control structure of the vehicle, reasonably designing an integrated chassis control algorithm and optimizing moment distribution of a dispatching motor.

The invention relates to automation conveying equipment, in particular to an automation conveying mechanism of a container. A visual guidance automatic skew detection type leather laser cutting system comprises a workbench, a laser labeling machine, an industrial camera, a feeding mechanism and a two-shaft robot; the feeding mechanism and the two-shaft robot are fixedly connected to the workbench, the laser labeling machine is located at the upper portion of the two-shaft robot, and the industrial camera is located at the upper portion of the feeding mechanism. Leather is wound on a winding disc and is pulled through the two-shaft robot to be fed, the pulled-out leather is cut off through the laser labeling machine according to the preset length, or various patterns are cut at the position of the cutting face, unmanned operation of leather cutting is achieved, and the high machining speed is obtained; and positional deviations of the leather are collected by the industrial camera and are corrected through the two-shaft robot, and the leather cutting quality is improved.

The invention provides a compound finite domain multiplier based on a heartbeat model. The compound finite domain multiplier based on the heartbeat model comprises an input port, a compound finite domain multiplier, a sub-domain multiplier, a sub-domain adder, a controller and an output port, wherein the input port is used for inputting an operand of compound finite domain, an irreducible polynomial selected on a sub-domain of the compound finite domain, an irreducible polynomial selected on the compound finite domain and a clock signal; the compound finite domain multiplier is used for multiplying the operand on the compound finite domain; the sub-domain multiplier and the sub-domain adder are respectively called by the compound finite domain multiplier for multiplying and adding the operand on the compound finite domain; the controller is in signaling connection with the input port and the compound finite domain multiplier and is used for controlling the compound finite domain multiplier; and the output port is in signaling connection with the controller and is used for outputting an operation result of the multiplication executed by the compound finite domain multiplier. According to the invention, a method based on the heartbeat model is adopted for performing multiplication on the compound finite domain; compared with the present multiplier, the multiplication on the compound finite domain has an obvious speed advantage; and therefore, the compound finite domain multiplier can be widely applied to the fields of mathematics and engineering.

The invention discloses a high-speed scanner which comprises a transparent flatbed used for placing files to be scanned and a pressing plate used for pressing the files to be scanned and arranged above the transparent flatbed. The high-speed scanner is characterized in that a cavity for light to pass is arranged below the transparent flatbed, and the side wall of the bottom of the cavity is provided with a digital camera body used for imaging once. The high-speed scanner has simple structure and high scanning speed; especially in color scanning, the scanning speed of the high-speed scanner is tens of times higher than a traditional flat scanner.

The invention provides a finite field multiplier based on a binary tree structure. The finite field multiplier comprises an input port used for inputting the operand a(x) and the operand b(x) of a finite field GF(2<n>), an output port used for outputting the multiplying result c(x) of the operand a(x) and the operand b(x), and a binary tree structure used for executing the multiplying of GF(2<n>) with the operand a(x) and the operand b(x). The binary tree structure comprises n+1 layers, from top to bottom, the layers numbered from 1 to n include a left binary tree and a right binary tree, and the bottommost layer is the (n+1)th layer; each node of the (n+1)th layer is connected with three nodes of the nth layer. Multiplying of the finite field is achieved with the binary tree structure, the structure is simple, and the finite field multiplier has obvious speed advantages in multiplying of GF(2<n>) compared with an existing finite field multiplier, and can be widely applied to various engineering fields.

The invention discloses a method for producing BOM (Bill of Material) memory model building based on XML (extensive markup language). According to the method, BOM aggregation relationship is redefined, BOM memory model building based on XML is innovatively proposed, the problems of high redundancy and low execution efficiency are solved, ideal experimental effects are achieved according to experimental verification, and new methods and thoughts are proposed in terms of BOM memory model building.

The invention discloses an equivalence detection method of a combinational logic circuit. The method comprises the steps that a logic coverage equivalence detection problem is divided into circuit including detection sub-problems by expanding a complementary minor concept, complementary minors of all product terms of one circuit expression to the other circuit expression are computed one by one, then whether or not the complementary minors are tautologies is judged on the basis of construction of Shannon structure charts of the complementary minors of all the product terms, and finally whetheror not a coverage equivalent relationship exists between two circuits is determined according to the judgment result of the tautologies. The equivalence detection method has the advantages that a logic function is subjected to decomposition and order reducing treatment by computing the complementary minors of the product terms, accordingly the coverage equivalence verification speed is increased,the operability and the detection efficiency are high, and the memory explosion problem cannot happen. The experimental structure indicates that the equivalence detection method is stable and effective, and the testing result of the circuits which are obtained through three algorithms integrated by EXPRESSO software indicates that the equivalence detection method has the remarkable speed advantage compared with two detection algorithms based on a truth table and a BDD.

The invention discloses a differentiated-temperature-heating temperature metal bath and an achievement method. The differentiated-temperature-heating temperature metal bath comprises a reaction module, a temperature control module, a reaction container movement device and a biochemical reaction signal detector. The biochemical reaction signal detector obtains a physical signal generated by the biochemical reaction of a reaction solution after detecting the reaction module. The differentiated-temperature-heating temperature metal bath and the achievement method have the beneficial effects that according to the differentiated-temperature-heating temperature metal bath, differentiated temperature heating is achieved by moving a reaction container, and therefore high and low temperature switching in the PCR nucleic acid amplification reaction process is more rapidly achieved, and the manner of achieving differentiated temperature heating by moving the reaction container will have obvious speed advantages compared with the manner of conducting temperature rise and fall through semiconductors themselves.

The invention provides a real-time identification method and a real-time identification system for a pesticide spraying sector angle of an unmanned aerial vehicle. The method comprises the steps of acquiring a real-time image which comprises a pesticide applying nozzle and a pesticide applying spraying sector; extracting a green component in the real-time image, and using the green component as an objective image; acquiring a central pixel of the objective image; according to the central pixel, acquiring a pixel a and a pixel b which are separated from the central pixel by m pixel distances in a vertical line which passes through the central pixel; acquiring a first horizontal line which passes through the pixel a and a second horizontal line that passes through the pixel b; acquiring a right upper side boundary point, a right lower side boundary point, a left upper boundary point and a left lower side boundary point; and acquiring the sector angle of the pesticide spraying sector according to the right upper boundary point, the right lower side boundary point, the left upper boundary point and the left lower side boundary point. The real-time identification method and the real-time identification system can realize quick identification for the pesticide spraying sector angle and furthermore satisfy a requirement for real-time performance.

The invention discloses a link layer processing method, which comprises the following steps: a link layer processing command and a flow of a processor are set into a random access memory (RAM) and a register which are arranged in a link layer processing device in advance, and then the link layer processing device is enabled; the processor Host sends a link layer command to the link layer controller; the link layer controller analyzes the command and starts a communication process with the slave device; the master device and the slave device perform link layer command interaction, and when a corresponding communication data packet is received, the link layer controller requests the link layer processing device for service interruption; an interrupt processor in the link layer processing device informs the main state machine to start interrupt state query; the interrupt is responded to according to a preset process; and after the protocol processing is finished, the link layer controllerinforms the Host through the HCI interface, and the interaction is finished. Command analysis of the link layer is processed, interrupt / command interaction can be completed without a processor, and alarge number of frequent interrupt requests and interrupt processing tasks are avoided.

The invention provides a single-machine minimized total tardiness scheduling method based on a distributed robust model and belongs to the technical field of production scheduling and internal and external production resource optimization. The method comprises the steps that a distributed robust optimization model for solving the single-machine scheduling problem is established firstly to obtain atarget function expression of the model; then, the distributed robust optimization model is converted into an integer second-order cone programming model; the converted model is solved, arrangement combinations of all workpiece machining sequences are expressed in a search tree in an enumeration way, the search tree is pruned through a branch-and-bound algorithm, and finally an optimal single-machine scheduling scheme of the minimized total tardiness is obtained. The method gives consideration to uncertain factors of a production environment, so that the model is a determined single-machine model more conforming to the actual production situation compared with an assumed production environment, and the obtained scheduling scheme can be better applied to actual production.

The invention provides a method for determining a pre-equalization matrix and a testing device. A radio communication tester having a base station emulator and a channel emulator is provided. A device under test having at least two branches is provided. An NxM multiple-input multiple-output (MIMO) connection is established between a radio communication tester and a device under test. The NxM MIMO connection comprises at least two channels. The per-branch reference signal received power measurements are continuously forwarded from the device under test to the radio communication tester. A pre-equalization matrix is determined by a radio communication tester, wherein the presence of per-branch reference signal received power measurements and additive white Gaussian noise (AWGN) is taken into account in calculating the pre-equalization matrix. The pre-equalization matrix compensates for crosstalk between the at least two channels and balances the branches such that the device under test receives equal power at its receiving antenna.

The invention provides a high-speed twin network target tracking method based on positioning perception, and belongs to the technical field of computer vision. The method mainly comprises the following steps: taking AlexNet as a feature extraction subnet, and carrying out a feature extraction task of a template image and a search image; in order to enhance the characterization capability of the features, the invention provides a context enhancement module which captures abundant target information from local and global levels, and simultaneously provides a new feature fusion strategy which fully combines the context information of different scale features; and finally, the regression loss of the network is calculated by using Distance-IoU loss, and a tracker is guided to select a more accurate bounding box. While a small number of parameters are increased, the method can guarantee a high tracking speed, and effectively improves the tracking performance of a twinning network tracker in a complex scene.

The invention discloses a composite finite field division device based on a cardiac model. The circuit comprises a finite field GF (2n) serial adder SADD1, wherein the SADD2 and the SADD3 are used forconverting the SADD2 into SAD3; a finite field GF (2n) parallel adder PADD, Finite field GF (2n) multipliers MUL1, MUL2, MUL3, MUL4, MUL5, MUL6 and MUL7, finite field GF (2n) constant multipliers CMUL1 and CMUL2, finite field GF (2n) squarer EXP and finite field GF (2n) inverters INV. Finite field division operation is achieved through the cardiac model, and compared with an existing divider, a division method for calculating GF ((2n) 2) has the obvious speed advantage and can be widely applied to various engineering fields.

A method for testing a device under test (12) by using a test system (10) is disclosed. The method comprises: generating a wideband modulated signal; forwarding the wideband modulated signal to an input (22) of a device under test (12); separating an electromagnetic wave reflected at the input by the directional element (16); forwarding the reflected electromagnetic wave to a vector signal analyzer (24); processing a reference signal associated with the wideband modulated signal; and determining a channel response by taking the reference signal and at least one scattering parameter of the device under test (12) into account, wherein the scattering parameter depends on the reflected electromagnetic wave. Further, the present disclosure relates to a test system (10).

The invention provides a real-time identification method and a real-time identification system for a pesticide spraying sector angle of an unmanned aerial vehicle. The method comprises the steps of acquiring a real-time image which comprises a pesticide applying nozzle and a pesticide applying spraying sector; extracting a green component in the real-time image, and using the green component as an objective image; acquiring a central pixel of the objective image; according to the central pixel, acquiring a pixel a and a pixel b which are separated from the central pixel by m pixel distances in a vertical line which passes through the central pixel; acquiring a first horizontal line which passes through the pixel a and a second horizontal line that passes through the pixel b; acquiring a right upper side boundary point, a right lower side boundary point, a left upper boundary point and a left lower side boundary point; and acquiring the sector angle of the pesticide spraying sector according to the right upper boundary point, the right lower side boundary point, the left upper boundary point and the left lower side boundary point. The real-time identification method and the real-time identification system can realize quick identification for the pesticide spraying sector angle and furthermore satisfy a requirement for real-time performance.

The present invention relates to the embedding and extracting method of wavelet chaotic semi-fragile digital water print. Original image is 2D two-stage wavelet analyzed to obtain {LL2, HL2, LH2, HH2, HL1, LH1, HH1}, where, the pixel grey of LL2 sublayer is mapped to form chaotic initial value and chaotically iterated for several times to form water print image. Then, based on the quantized medium frequency coefficients { HL2, LH2, HH2}, and by means of the extreme sensitivity of chaos to the initial value and the excellent temporal frequency characteristic of wavelet, the distortion to water print image is precisely located. Via flexible regulation of quantized parameters, the robustness against baleful attack is obtained. The method of the present invention has stable performance, fast realizing speed, strong robustness to baleful distortion and excellent distortion locating and proving capacity.

The invention discloses an equivalence detection method of a combinational logic circuit. The method comprises the steps that a logic coverage equivalence detection problem is divided into circuit including detection sub-problems by expanding a complementary minor concept, complementary minors of all product terms of one circuit expression to the other circuit expression are computed one by one, then whether or not the complementary minors are tautologies is judged on the basis of construction of Shannon structure charts of the complementary minors of all the product terms, and finally whetheror not a coverage equivalent relationship exists between two circuits is determined according to the judgment result of the tautologies. The equivalence detection method has the advantages that a logic function is subjected to decomposition and order reducing treatment by computing the complementary minors of the product terms, accordingly the coverage equivalence verification speed is increased,the operability and the detection efficiency are high, and the memory explosion problem cannot happen. The experimental structure indicates that the equivalence detection method is stable and effective, and the testing result of the circuits which are obtained through three algorithms integrated by EXPRESSO software indicates that the equivalence detection method has the remarkable speed advantage compared with two detection algorithms based on a truth table and a BDD.

The invention provides a cleaning device for a titanium-magnesium alloy door panel. In order to prevent dust particles from scattering, the device is provided with a sealed cabin on the base, and a grinding module is placed in the sealed cabin, thereby avoiding the dust generated by grinding. polluted environment. On this basis, the present invention is provided with a blower and a dust collection bag, under the action of the blower, a positive air pressure is generated in the airtight chamber, and when the airflow flows through the dust collection bag, it plays a role of dust collection, thereby alleviating the sealing effect over time. Excessive accumulation of dust in the cabin affects the operation of the device. In order to further avoid the problem of dust adhering to the motor and the wall of the sealed cabin, the present invention is provided with an ultrasonic oscillator on the inner wall of the sealed cabin, and a warehouse wall vibrator on the motor, so as to avoid dust adhesion by using ultrasonic waves and mechanical vibration. Ensure the normal operation of the device.

A remote sensing image near-lossless compression hardware realization method based on improved JPEG-LS algorithm comprises the steps (1) directly calculating to obtain a pixel reconstruction value by the pixel actual value of an input image and accomplishing the calculation in a single clock period by using a formula Rx=int[Ix / (2Near+1)]*(2Near+1), in the formula, the Rx and the Ix are respectively the pixel reconstruction value and actual value, the int is rounding operation, and the Near is a compression ratio control factor; (2) calculating a context environmental variable Q according to the obtained pixel reconstruction value, if Q is equal to 0, performing run coding, otherwise, going to a step (3) of performing conventional coding; (3) calculating the predictive value of the currentpixel by the pixel reconstruction value according to the geometric position relation between the current pixel and the adjacent pixel; (4) calculating the residual value between the predictive value and actual value of the current pixel; and (5) performing Golomb coding after quantising the obtained residual value, and synchronously updating the parameter variable corresponding to the context environmental variable Q using the quantisation result.