49results about How to "Improve recognition stability" patented technology

The invention provides an automated guided vehicle navigation method based on two-dimension code positioning and relates to the technical field of navigation of automated vehicles. The automated guided vehicle navigation method based on two-dimension code positioning solves the technical problems that according to an existing method, deployment is not convenient, investment cost is high, and paths are not flexible. According to the automated guided vehicle navigation method based on two-dimension code positioning, a two-dimension code label array which is arranged in a mesh mode is attached to the ground of the site where an automated guided vehicle is applied, two-dimension code labels on the ground are obtained through a camera and are recognized when the automated guided vehicle advances, the automated guided vehicle is controlled to advance according to the position information contained in the two-dimension code labels, and thus navigation of the automated guided vehicle is achieved. The automated guided vehicle navigation method based on two-dimension code positioning has the advantages that deployment is convenient, investment cost is low, and the paths are flexible.

The invention discloses an intelligent wind turbine generator working condition identification method which is used for identifying working conditions of a wind turbine generator comprising a plurality of subsystems. The method includes the steps of obtaining a plurality of parameters of the wind turbine generator; dividing the parameters into global parameters and local parameters; extracting eigen values of the plurality of parameters; and obtaining an eigen value vector of each parameter according to the eigen value of each parameter; and classifying the eigen value vectors of the plurality of parameters into global working conditions and subsystem working conditions through a first layer self-organizing map neural network and a second layer self-organizing map neural network to obtain a working condition identification result of the wind turbine generator. The plurality of parameters of the wind turbine generator are adopted to analyze the wind turbine generator working conditions, the wind turbine generator working conditions are divided into the plurality of global working conditions and subsystem working conditions, wind turbine generator working condition identification stability can be improved, and monitoring precision for wind turbine generator running is improved.

The invention relates to an underwater moving target identification method. The underwater moving target identification method includes the following steps: conducting preprocessing to target signals received by active sonar, conducting self-mixing (SM) time frequency processing to the target signals after being preprocessed, and extracting a micro-doppler spectrum; extracting intra-frame micro-doppler features and interframe micro-doppler features according to the micro-doppler spectrum; and classifying underwater moving targets based on the intra-frame micro-doppler features and the interframe micro-doppler features. The underwater moving target identification method extracts micro-doppler distribution feathers through echo waveform of underwater moving small target of an active high frequency sonar. A SM algorithm can produce self-correlation items which are identical with west data (WD), mutual disturbance terms cannot exist, the problem that micro-doppler variation restricted by linear frequency modulation (LFM) signal ambiguity function can not be achieved is solved, classification of underwater moving small target of active high frequency sonar single beam echo waveform is achieved, influences of a random channel are reduced through statistics stability of the interframe micro-doppler features, and identification stability is improved.

The invention relates to an information technology automatic control field, the method is made up of road sign designing method, road sign position, road sign incriminating and navigating device, and road sign incriminating and control flow. The road sign designing method is a gray mode road sign, which is made up of two long black bar with the same width, four transverse long black bar with the same width and the middle numbers, the printed road sign is pasted on the plane vertically to the ground, such as the wall or side surface of computer desk, the height is similar to the camera height of robot, the incriminating and navigating device is only robot with a single camera, finally, it is accomplished road sign incrimination and control flow. The flow carries on binarization to the 256 gray images, then it is scanned and detected, finally the road sign number can be incriminated; the mode is simple, and the cost is low.

The invention discloses a mobile robot positioning method based on a laser scanner and strong reflective signs. The method comprises that steps that a mobile robot environment monitoring system composed of the multi-line laser scanner and the strong reflective signs is established, and the strong reflective signs are recognized and relatively positioned through environment light intensity data and distance data obtained by the multi-line laser scanner. When the initial position of the robot is not known, distance and gradient information of pairs of connected lines of the reflective signs is calculated, therefore global positioning of the reflective signs is achieved through data matching, and finally position information acquisition of the robot under the global environment is achieved through the relative position relation between the robot and the reflective signs. The mobile robot positioning method has the beneficial effects that the defects that perceptual information is excessively simple when only distance or direction data are adopted, and less target characteristics are obtained are overcome, and recognition stability is improved.

The invention discloses a positioning and tracking system of an underground / roadway moving object, comprising underground positioning and tracking systems distributed under a detected mine and ground monitoring systems distributed on the ground; two-way communication is conducted between the underground positioning and tracking system and the ground monitoring system; the underground positioning and tracking system comprises a plurality of wireless radio-frequency labels, a plurality of wireless signal receiving and transmitting devices and a plurality of card readers which are respectively carried by underground workers, are internally stored with the information of the identity and engaged job category information of a cardholder, can read the stored information in the wireless radio-frequency labels within respective signal coverage area in real time, and synchronously upload the read information and the internally-stored information to the card reader of the receiving and transmitting device; and the wireless radio-frequency labels are distributed on a coal mining work surface or a coal mining roadway. The positioning and tracking system has reasonable design, convenient installation and distribution, low cost, complete function and good use effect, and also improves the coverage area correspondingly while saving the cost.

The invention belongs to the field of image target recognition and provides a remote sensing ship identification method based on dense feature fusion and pixel-level attention, and aims to solve the problems that a classical neural network easily identifies a plurality of dense targets as one target under a remote sensing image ship target identification task, a large number of small targets are missed, boundary frames are easy to overlap and the like. According to the main scheme, data set division is carried out on a remote sensing image data set to obtain a training set and a test set, anddata enhancement of the training set is carried out. RGB three-channel average values r < mean >, g < mean > and b < mean > of the original remote sensing image data set are calculated and the RGB three-channel values of the images are correspondingly subtracted in the expanded data set from the r < mean >, g < mean > and b < mean >; the obtained data set is input into an improved Faster RCNN network to be trained, core modules of the network are a dense feature fusion network and a pixel-level attention network, and the network outputs candidate rotation boxes and category scores of the candidate rotation boxes; and skew IOU-based rotating frame non-maximum suppression is carried out on the obtained result to obtain an identification result of the ship target in the remote sensing image.

The invention discloses an active RFID identification system and a control method thereof for accurately reading a boundary. The active RFID identification system consists of an active RFID card reader and an active RFID card, wherein the active RFID card reader comprises a low-frequency transmitting device used for continuously transmitting low-frequency trigger signals, and a high-frequency receiving device used for receiving a confirmation signal transmitted by the active RFID card; and the active RFID card comprises a low-frequency receiving device used for receiving the low-frequency trigger signals transmitted by the low-frequency transmitting device of the active RFID card reader and modulating coded information of the low-frequency trigger signals at the same time, and a high-frequency transmitting device used for receiving a confirmation signal transmitted by a logic control circuit and transmitting the confirmation signal to the receiving device of the active RFID card reader. The invention combines advantages of the low-frequency trigger signals and a high-frequency receiving function, so the card reading distance can be accurately controlled.

The invention relates to a band service optical fiber low-frequency modulation generation device on the basis of macrobend loss, which comprises a motor, a transmission assembly, a bending clamping assembly and a base. The transmission assembly comprises a cylindrical cam, a rolling bearing, a push rod, a pull rod and a tension spring; the tension spring is acted on the push rod and enables the rolling bearing at the tail part of the push rod to be pressed against the inclined plane of the cylindrical cam; the bending clamping assembly is arranged at the front end of the push rod and comprises a four-bar mechanism; the four-bar mechanism clamps an optical fiber by two pairs of clamping pillars on the four-bar mechanism; a wrapping post is arranged between the two pairs of clamping pillars; the optical fiber can be bent at the position of the wrapping post; the motor drives the cylindrical cam to synchronously rotate, the inclined plane of the cylindrical cam drives the push rod to move back and forth by matching with effects of the tension spring, the push rod drives the four-bar mechanism to carry out reciprocating deformation and the four-bar mechanism drives the optical fiber to carry out reciprocating bending motion under the specific frequency. According to the and service optical fiber low-frequency modulation generation device, the band service optical fiber can be subjected to macro bending under the specific frequency and at a variable bending angle and low-loss specific-frequency power modulation is generated in the optical fiber.

According to an EOF-based unmanned aerial vehicle signal identification system and method provided by the invention, the problem that similar unmanned aerial vehicle signals are difficult to identifyis solved, the identification effect of unmanned aerial vehicles with similar signals can be effectively improved, and better support is provided for unmanned aerial vehicle identification managementand control. The method provided by the invention comprises the following steps: 1) firstly, carrying out preliminary identification on a received unmanned aerial vehicle signal through a characteristic parameter matching method; 2) carrying out FFT (Fast Fourier Transform) processing on the unmanned aerial vehicle signal which cannot be recognized through characteristic parameter matching, and then generating a corresponding time-frequency analysis graph; 3) adopting an empirical orthogonal function analysis method to decompose a video analysis matrix corresponding to the video analysis graphinto an EOF matrix reflecting signal common characteristics and a characteristic coefficient matrix reflecting signal difference characteristics; and 4) performing feature data extraction and BP neural network training on the feature coefficient matrix, constructing a corresponding unmanned aerial vehicle type classifier, and outputting an identification result.

The invention relates to the technical field of freight train detection, in particular to a railway wagon lower pull rod falling fault image recognition method, and aims to solve the problem of low detection accuracy in a manual-based detection mode in the prior art, and the method comprises the steps of 1, obtaining a to-be-recognized wagon image, and constructing a sample data set; step 2, carrying out data amplification processing on the sample data and marking a sample data set; 3, constructing a training data set according to the data obtained in the step 2; 4, constructing a MultiResUNetmodel, and training the MultiResUNet model to obtain a trained deep learning model; 5, inputting a to-be-recognized image into the trained deep learning model to obtain a segmented image of the pull-down rod; and 6, judging whether the pull-down rod is in a falling state or not by utilizing the segmented image of the pull-down rod. Manual labor is replaced by artificial intelligence, and the accuracy and efficiency of a detection result are improved while manpower is saved.

The invention discloses a visual identification method and device for single-category unstacking, equipment and a medium, and belongs to the field of machine vision, and the visual identification method comprises the steps: carrying out the filtering and down-sampling of an original 3D point cloud of a to-be-unstacked object, and obtaining a target point cloud; performing boundary extraction on the target point cloud to obtain boundary point cloud, wherein the boundary point cloud only comprises contour points; performing 3D matching on a preset template point cloud and the boundary point cloud, and segmenting out each workpiece contour point cloud matched with the preset template point cloud from the boundary point cloud, the preset template point cloud being a contour point cloud consistent with each workpiece type in the to-be-unstacked object; and obtaining workpiece poses corresponding to the workpiece contour point clouds obtained through matching, and performing unstacking according to height information in the workpiece poses. According to the visual identification method, 3D contour matching is carried out on the filtered workpiece point cloud and template point cloud of which the boundaries are extracted, so that the interference of a large amount of plane point clouds is reduced, and the matching precision is improved, and the calculation speed is increased.

The invention discloses a multimedia volume adjusting method and system. The method comprises the steps of obtaining a depth image of a user posture; conducting image division of a hand target area on the depth image; obtaining a divided target area image; using a Sobel operator to conduct gesture edge profile detection according to the target area image to extract edge parameter features; using a DAG-SVMs classifier to obtain a gesture classification result according to the edge parameter features; adjusting the volume level of a multimedia according to the gesture classification result, wherein different gesture classifications correspond to different volume levels. According to the multimedia volume adjusting method and system, the control over the volume of the multimedia itself can be achieved through gesture operations simply by making multimedia equipment connected to one piece of externally-arranged depth image acquisition equipment, in the control process, a common digital gesture in life is adopted, and the external equipment is simple, high in operability and high in convenience in the whole control process.

The invention discloses a traffic-light driverless auxiliary device and method based on the Faster RCNN, and the device mainly comprises a main control Raspberry Pi 3B+, a wide angle camera, a buzzerprompt module, and a liquid crystal display screen module. The main control Raspberry Pi 3B+ is provided with a deep learning frame TensorFlow, and operates an RCNN (regional convolutional neural network) model. The wide angle camera collects a current road image in real time, determines the specific position of traffic lights through the deep learning model, and obtains a bounding rectangle. An ROI of the traffic lights is cut through OpenCV according to the bounding rectangle, so the calculation burden is smaller. After conventional filtering and other preprocessing operations, an image is converted into a Lab color space form a BGR color space, and channel a is independently extracted, and the traffic lights are recognized according to the ratio of the area of a connected domain. A final result will be prompted through the buzzer module and the display screen module. The device and method facilitate the reduction of traffic accidents, and improves the safety coefficient of driving.

The invention provides an identity card element extraction method and device, and the method comprises the steps of correcting an obtained identity card image, and obtaining a target image matched with a pre-stored standard identity card template; determining a character block detection box of the target image by performing character block detection on the target image; extracting identity card elements based on the character block detection box. The invention can solve a problem that a mode of character line segmentation based on binary image projection is greatly affected by background interference in the prior art. Through correction, the identity card image with rotation and perspective can also have a good recognition effect, identity card elements are extracted based on the characterblock detection box, high-quality text line positioning information is provided for text recognition, and the recognition stability is improved.

The invention relates to a power plant instrument image detection and recognition method and device, and the method comprises the steps: collecting an original image of a power plant instrument, carrying out the image preprocessing, and extracting an instrument dial region in the original image to form a target image; processing the target image through an improved Canny detection algorithm; carrying out non-maximum suppression on the gradient amplitude; obtaining an upper threshold value and a lower threshold value of the image after non-maximum suppression by adopting a self-adaptive threshold value confirmation algorithm; determining the boundary of the image through a double-threshold algorithm according to the upper threshold and the lower threshold; detecting straight lines and circles in the binary boundary image through a Hough change algorithm; and obtaining the range reading of the instrument according to the straight line and the circle in the binary boundary image through apointer instrument identification method. Compared with the prior art, the instrument image is identified by combining the Canny detection algorithm and the Hough transformation algorithm, and the pointer type power plant instrument is rapidly and accurately identified; meanwhile, the recognition stability is high.

The invention discloses a polarized light time domain reflectometer based on three-polarization-state time division multiplexing. The polarized light time domain reflectometer comprises a light source, a polarization state generator, a pulse modulator, a circulator, a sensing optical fiber, a dual-polarization state detection device, a photoelectric detector, a signal acquisition card and a processor, wherein the processor is connected with the polarization state generator; the polarization state output by the polarization state generator is controlled to be switched among the three polarization states perpendicular to one another; and polarization detection is achieved through the double-polarization-state detection device, so that under the condition that depolarization is not considered, the phenomenon of traditional POTDR signal fading is completely eliminated, and the effect of improving system positioning and event recognition stability is achieved.

The present invention relates to a non-contact identity verification device and system, and a method thereof, comprising: a physiological characteristic detector, an identification unit, and a transmission unit. The physiological characteristic detector detects a physiological signal of a user. The identification unit compares the physiological signal of a user with a physiological characteristic check value and generates a comparison information. The transmission unit transmits the comparison information. Through a verification technology of the present invention, a user can be verified rapidly and conveniently, recognition stability is very high, and personal data of the user can be adequately protected.

An automatic guided vehicle navigation method based on QR code positioning relates to the technical field of automatic vehicle navigation. It solves the technical problems of inconvenient deployment, high investment cost and inflexible routes of existing methods. This method places a grid-like array of QR code labels on the ground where the automated guided vehicle is used. When the automated guided vehicle travels, it uses a camera to obtain the QR code labels on the ground and identifies them, and controls them based on the location information in the QR code labels. The automatic guided vehicle travels to realize the navigation of the automatic guided vehicle. The method provided by the invention is easy to deploy, has low investment cost and is flexible in route.

The invention discloses a polarized light time domain reflectometer based on dual polarization state time division multiplexing. The polarized light time domain reflectometer comprises a light source,a phase modulator, a pulse modulator, a circulator, a sensing optical fiber, a full polarization state detection device, a photoelectric detector, a signal acquisition card and a processor, wherein an output port of the phase modulator is connected with an input port of the pulse modulator through an optical fiber, and the phase modulator is used for enabling the polarization state of light output by the phase modulator to be perpendicular to each other on the Bonding ball; and the full polarization state detection device is connected with the photoelectric detector through an optical fiber,and the full polarization state detection device is used for realizing detection of three polarization states which are perpendicular to one another. According to the polarized light time domain reflectometer, dual-polarization-state time division multiplexing full-polarization-state detection is used, so that the phenomenon of traditional POTDR signal fading is completely eliminated under the condition that depolarization is not considered, and the effect of improving system positioning and event recognition stability is achieved.

The invention discloses a mobile robot positioning method based on a laser scanner and strong reflective signs. The method comprises that steps that a mobile robot environment monitoring system composed of the multi-line laser scanner and the strong reflective signs is established, and the strong reflective signs are recognized and relatively positioned through environment light intensity data and distance data obtained by the multi-line laser scanner. When the initial position of the robot is not known, distance and gradient information of pairs of connected lines of the reflective signs is calculated, therefore global positioning of the reflective signs is achieved through data matching, and finally position information acquisition of the robot under the global environment is achieved through the relative position relation between the robot and the reflective signs. The mobile robot positioning method has the beneficial effects that the defects that perceptual information is excessively simple when only distance or direction data are adopted, and less target characteristics are obtained are overcome, and recognition stability is improved.

The invention relates to an information technology automatic control field, the method is made up of road sign designing method, road sign position, road sign incriminating and navigating device, and road sign incriminating and control flow. The road sign designing method is a gray mode road sign, which is made up of two long black bar with the same width, four transverse long black bar with the same width and the middle numbers, the printed road sign is pasted on the plane vertically to the ground, such as the wall or side surface of computer desk, the height is similar to the camera height of robot, the incriminating and navigating device is only robot with a single camera, finally, it is accomplished road sign incrimination and control flow. The flow carries on binarization to the 256 gray images, then it is scanned and detected, finally the road sign number can be incriminated; the mode is simple, and the cost is low.

The embodiment of the invention discloses an image processing method, device and terminal and a computer readable storage medium. The method comprises the steps: obtaining a processing region, comprising a specific character string, in a to-be-processed image, and performing the contrast enhancement processing of the processing region to obtain a first processing image; obtaining the colors of pixels in the first processing image, and judging whether there is a damage in a first processing image or not according to the colors of the pixels; removing the damage in the first processing image according to a preset processing model when there is the damage in the first processing image, and performing the character recognition according to the image after the damage is removed. The method canremove the interference caused by the damage in a character recognition process, improves the character recognition accuracy, can resist various types of damages in the actual samples, and is high inrecognition stability.

The invention discloses a double-stage time sequence action detection method and device, equipment and a medium. The method comprises the steps of obtaining video information features; finding according to video information features, potential action starting and ending moments; combining the starting time and the ending time into a candidate frame; and calibrating the boundary of the candidate box, and judging the content of the candidate box to obtain an action category. The two-stage time sequence action detection method, device and equipment and the medium have the advantages of being high in recognition precision, good in recognition stability, good in robustness and the like.

The invention provides an OCR identification method based on model fusion. According to the method, a plurality of recognition models are adopted to build a deep learning neural network model, the recognition models train character positioning data respectively until the models converge to obtain a pre-output recognition result, and the pre-output recognition result is fused and optimized to output a final recognition result. The multi-model fusion OCR recognition problem is solved, and compared with the traditional OCR recognition method, the deep learning multi-model fusion OCR recognition technology is adopted, so that positioning is accurate, the recognition accuracy is high, and the model recognition effect is more stable.

The invention discloses a network flow control method and control system based on DFI and DPI, relates to the technical field of network data transmission, and is used to reduce the identification burden of network flow and improve the identification stability and identification efficiency of network flow. The network traffic control method includes: obtaining service identification information of network traffic, and formulating a traffic identification control strategy according to the service identification information; configuring a sample module according to the traffic identification control policy, and controlling a DFI identification unit and / or a DPI identification unit to perform network traffic Detection and identification; the DFI identification unit and / or DPI identification unit obtains the network traffic identification result, and stores the network traffic identification result to the data exchange module; the data exchange module feeds back the network traffic identification result to the traffic identification control module. The network traffic control method and control system based on DFI and DPI provided by the present invention are used for network traffic identification.

The invention discloses a polarized light time domain reflectometer based on three-polarization-state time division multiplexing. The polarized light time domain reflectometer comprises a light source, a polarization state generator, a pulse modulator, a circulator, a sensing optical fiber, a dual-polarization state detection device, a photoelectric detector, a signal acquisition card and a processor, wherein the processor is connected with the polarization state generator; the polarization state output by the polarization state generator is controlled to be switched among the three polarization states perpendicular to one another; and polarization detection is achieved through the double-polarization-state detection device, so that under the condition that depolarization is not considered, the phenomenon of traditional POTDR signal fading is completely eliminated, and the effect of improving system positioning and event recognition stability is achieved.

The invention provides a serial number switching control circuit, a working method of the serial number switching control circuit, a consumable chip, a working method of the consumable chip, a consumable container and printing equipment. The serial number switching control circuit comprises a main control circuit, a diode circuit group, a power supply end, a detection resistor and a voltage detection circuit; the diode circuit group is electrically connected with the power supply end through the detection resistor, and the voltage detection circuit obtains the detection voltage of the detection resistor and sends a voltage feedback signal to the main control circuit; the diode circuit group comprises multiple paths of diode circuits which are connected in parallel; a plurality of serial numbers are stored in the main control circuit, and one serial number corresponds to one diode circuit. According to the serial number switching control circuit, the occupied storage space can be reduced, and the serial number identification stability is improved.

The invention provides an electronic tag. The electronic tag includes: an antenna, wherein the antenna comprises a feeder line and a load, the feeder line comprises a feeder line core with a feeder line core pin and a shielding layer arranged on the feeder line core, the load comprises a load core with a load core pin and a load ground, the load ground is arranged on the periphery of the load coreto insulate the feeder line core from the load core, and the load ground is connected with the shielding layer on the feeder line core through a conducting strip; And a label, wherein the label comprises a label body, the label body is provided with a first pin and a second pin, and the center of the label body is provided with an RFID chip. When the feed line core pin is in contact with the first pin and the load core pin is in contact with the second pin, the feed line and the load form a closed loop through the label. The electronic tag does not relate to electromagnetic wave conversion, the peripheral electronic tag cannot be identified under the condition that the antenna is not in contact with the tag, the identification precision is high, and the identification stability is high.