293results about How to "Improve the recognition success rate" patented technology

The invention discloses an unmanned management method of a parking lot and an unmanned management system of the parking lot. The unmanned management method of the parking lot comprises the following steps: S1, when a vehicle is detected that the vehicle enters a sensing range of an exit / entrance of the parking lot, triggering a photographic device to be started so that a two-dimensional code or bar code image, which is displayed on a mobile terminal screen in the vehicle and which carries vehicle owner identity information, is remotely scanned; S2, decoding the scanned two-dimensional code or bar code image, extracting the vehicle owner identity information carried by the two-dimensional code or bar code image, uploading the vehicle owner identity information to a cloud server, and verifying the vehicle owner identity information through the cloud server; and, S3, when verification of the vehicle owner identity information by the cloud server at the cloud server is passed, receiving a switching-off instruction issued by the cloud server, and controlling a barrier gate of the exit / entrance of the parking lot to arise so that a vehicle owner conveniently drives in or out of the parking lot. Accurate identification of the vehicle owner identity information can be achieved, and the vehicle owner can be prevented from using a real parking card.

The invention relates to a two-dimensional code scanning and recognizing method and a mobile terminal. The two-dimensional code scanning and recognizing method comprises the steps of scanning a two-dimensional code and performing recognition; if recognition for the two-dimensional code fails, acquiring an original image of the two-dimensional code; if the original image meets a preset condition, adjusting a photographing assembly of the mobile terminal according to an adjustment strategy corresponding to the preset condition, and acquiring a target image of the two-dimensional code; and performing recognition on the target image. According to the embodiment of the invention, automatic intelligent adjustment is performed on the photographing assembly in the two-dimensional code scanning process, so that the image quality of two-dimensional code scanning is improved, and the success rate of two-dimensional code recognition is improved.

The invention disclosed an equipment control method, device and system. The method comprises the following steps that: a server receives voice information sent by a control end; the server performs semantic recognition on the voice information to determine a control instruction corresponding to the voice information; and the server sends the control instruction to controlled equipment so as to make the controlled equipment execute corresponding operation according to the control instruction. According to the embodiment of the invention, the control end sends the voice information to the server, and the server is utilized to perform on semantic recognition on the voice information to determine the control instruction, and therefore, limitation on inputted voice information can be reduced; the server can not only recognize voice information consistent with a preset control instruction, and can also recognize daily colloquial voice information, and therefore, flexibility of recognition on voice information can be improved, and the success rate of the recognition can be improved, and user experience can be enhanced.

The invention provides a bar code recognition method and device. The method comprises the following steps of: performing binaryzation on a bar code image acquired by a user terminal; transversely projecting the bar code image subjected to the binaryzation, and positioning a character zone according to the distribution condition of transverse projection values; longitudinally projecting the character zone, and positioning each character according to the distribution condition of longitudinal projection values; and performing character recognition on each character, and combining the character recognition result to obtain a bar code recognition result. According to the invention, the character recognition technology is introduced into bar code recognition, the bar code recognition result is obtained in a mode of recognizing the characters below the bar code instead of a decoding mode of calculating the widths of black and white bars, and the success rate of bar code recognition is improved at the user terminal with limited shooting conditions.

The invention provides a visual guiding-based robot workpiece grabbing method, and belongs to the field of robot visual recognition and positioning. The method comprises the following steps of (1) establishing templates of target workpieces in advance, collecting images of the target workpieces, and acquiring edge feature images of the templates and edge feature images of the target workpieces; (2) acquiring the deflection angle of each target workpiece relative to the corresponding template by utilizing the minimum enclosing rectangle of the edge feature image of the target workpiece, and establishing a compensation template set by utilizing the deflection angles and the edge feature images of the templates; (3) matching the edge feature images of the target workpieces and the compensation template set, and recognizing the target workpieces and acquiring the center coordinates of the target workpieces under a visual coordinate system; (4) converting the visual coordinate system to a robot user coordinate system; and (5) selecting a reference point, and uniquely identifying the same target workpiece by utilizing the time and the position to complete positioning grabbing. The targetworkpieces can be accurately recognized by utilizing the method.

An embodiment of the invention provides an application running device and a terminal device. The application running device comprises a scanning module, a starting operation detection module, a first safety detection module, a starting module, an event detection module and a second safety detection module; the scanning module is used for scanning a specified type of application in the current operation system; the starting operation detection module is used for detecting the starting operation of the specified type of application; the first safety detection module is used for performing the first safety detection on the running environment of the application before the application is started; the starting module is used for starting the application after a result of the first safety detection indicates that the running environment is safe; the event detection module is used for detecting a preset event in the running process of the application; the second safety detection module is used for triggering the second safety detection on the preset event in the application. According to the application running device and the terminal device, the safety of the running environment when the application is started is effectively ensured and the safety of the application in the running process is effectively ensured.

The invention provides a fingerprint information processing method, which comprises the following steps of uniformly carrying out normalization processing on images and carrying out image segmentation; carrying out image enhancement through image smoothing and median filtering; converting gray level images into binaryzed images by utilizing a dynamic threshold value; carrying out fining processing on the images, and deleting pseudo characteristic points. The invention provides the fingerprint information processing method; a user is enabled to intuitively know image quality; moreover, the identification success rate is improved; the fingerprint information processing method is easy to realize; moreover, the execution efficiency is higher.

The invention belongs the technical field of align and discloses substrate align marks, a manufacturing method of the substrate align marks and a substrate. Patterns of the two align marks are manufactured on structures of different layers of the substrate, when one align mark can not be identified when align is conducted due to the fact that a coating film is uneven in the forming process of the one align mark, and the align can be further conducted by identification of the other align mark, the identification success rate of the align marks is improved, and therefore the defective percentage due to the fact that the align marks can not be identified is greatly reduced in the manufacturing process of the substrate.

The invention discloses an injection attack detection method based on a gating loop unit and an attention mechanism, and the method is characterized in that the method comprises the following steps: step 1, extracting a statement related to a database in the Internet as a sample; preprocessing each sample, and obtaining a category label of each sample; step 2, inputting a result after sample preprocessing into an injection attack detection model, and predicting a category label corresponding to the result; wherein the injection attack detection model comprises an activation layer, a gating cycle unit layer, an attention mechanism layer and an output layer which are connected in sequence; step 3, training an injection attack detection model based on the sample data; and step 4, for a to-be-detected statement, preprocessing the to-be-detected statement, and inputting a preprocessing result into the trained injection attack detection model to obtain a category label of the to-be-detectedstatement, thereby judging whether the to-be-detected statement is an injection attack statement or not. The method is high in accuracy.

The invention discloses a face identification method and device. The method comprises that to-be-identified face information of a present frame in a video stream and a feature distance value, of reference face information, stored in a face feature library are obtained; whether the to-be-identified face information matches the reference face information is determined according to the relation between the feature distance value and a preset threshold; and if the to-be-identified face information matches the reference face information, the preset threshold is improved and serves as a new preset threshold. According to the invention, the dynamic threshold related to a matching result of the to-be-identified face information of the last frame is used, the to-be-identified face information and the feature distance value of the reference face information are prevented from being influenced by change of the surrounding and identification mistakes, and the success rate of face identification is improved.

The invention provides an authentication method based on fingerprints. The method comprises the step of: collecting the fingerprints by utilizing a fingerprint identification device. When a quality evaluation parameter of a collected fingerprint image is less than a preset quality threshold, a user is prompted, and the fingerprints are recollected. The invention provides the authentication method based on the fingerprints, so that the user visually knows about the image quality, the identification success rate is increased, the implementation is easy, and the execution efficiency is high.

The present invention provides a human body re-recognition method. A human torso is used as a first-layer image region of human body scale space division, and each subsequent level of human body scale space division is multi-level human body scale space division performed on a human body image in an equal-area division manner based on a previous level of image region division, to acquire a multi-level human body image region; and a plurality of color channels are chosen, spatial histogram calculation is performed on each color channel according to each level of human body image region obtained through the multi-level human body scale space division, and then the plurality of color channels are cascaded to form a multi-channel spatial histogram. The method provided by the present invention solves the problem that feature description and feature matching in the prior art have defects, and the method has a high recognition success rate and a low false alarm rate.

The invention discloses a whole-process automatic attendance system used in a classroom. The whole-process automatic attendance system comprises a collecting system, an analyzing system and a statistical system, wherein the collecting system comprises cameras; four cameras are arranged in total and are respectively arranged at left upper corner and right upper corner on the front side of the classroom, above the left wall of the classroom and above the right wall of the classroom; the analyzing system comprises a database, in which the face standard data of attendance people are stored in advance, and is used for comparing the real-time images collected by the cameras with the pre-stored data in the database and then sending a comparison result to the statistical system; the statistical system is used for generating a report for the attendance people according to the comparison result. The invention has the beneficial effects that the accuracy is higher and the whole-process attendance recording for people in the classroom can be accurately performed.

An embodiment of the invention provides an application running method. The application running method comprises scanning a specified type of application in the current operation system; detecting the starting operation of the specified type of application; performing the first safety detection on the running environment of the application before the application is started; starting the application after a result of the first safety detection indicates that the running environment is safe; detecting a preset event in the running process of the application and triggering the second safety detection on the preset event in the application. According to the application running method, the safety of the running environment when the application is started is effectively ensured, the safety of the application in the running process is effectively ensured, the comprehensive safety protection is provided for the application, and the safe starting and running of the application can be achieved.

An embodiment of the invention provides an application running system. The application running system comprises a driving assembly and a process assembly; the process assembly is used for scanning a specified type of application in the current operation system; the driving assembly and or the process assembly are or is used for detecting the starting operation of the specified type of application; the driving assembly and the process assembly are used for performing the first safety detection on the running environment of the application before the application is started; the driving assembly is used for starting the application after a result of the first safety detection indicates that the running environment is safe; the driving assembly is used for detecting a preset event in the running process of the application; the driving assembly and the process assembly are used for triggering the second safety detection on the preset event in the application. According to the application running system, the comprehensive safety protection is provided for the application and the safe starting and running of the application can be achieved.

The invention belongs to the field of three-dimensional staring point recognition and computer vision, and particularly discloses a pose recognition method, device and system for an object of interestto human eyes. The method comprises the steps of utilizing a left eye camera and a right eye camera to recognize the centers of the left and right pupils of a user respectively so as to extract the information of the human eyes; mapping the identified left and right pupil centers into a foreground left camera to obtain the two-dimensional staring points; extracting an object anchor frame in the foreground left camera by utilizing a target identification and tracking algorithm, and then determining an object of interest of the user according to the position relation between the two-dimensionalstaring point and the object anchor frame; performing three-dimensional reconstruction and attitude estimation on the object of interest of the user to obtain a pose of the object of interest in theforeground left camera; and converting the pose of the object of interest in the foreground left camera into a world coordinate system so as to determine the pose of the object of interest of the user. The method, device and system can identify the object of interest of the user and estimate the pose of the object, and has the advantages of high identification accuracy, high pose estimation precision and the like.

The invention provides a request behavior characteristic based user identification method, comprising the following steps: firstly, receiving a request message sent from a user terminal; then, decoding and analyzing the request message to obtain a user identification code; according to the user identification code, checking the historical request record of the user; according to the historical request record, calculating the eigenvalues of the user's request behavior wherein the request behavior includes the request frequency characteristics and / or the corresponding relationship characteristic; and finally, determining whether the eigenvalues of the request behavior are larger than a threshold value of the eigenvalues and based on the determination result, identifying whether the user is a normal user or a non-normal user. Compared with the CAPTCHA technology that is applied to the user's front end, the method of the invention is applied to the back end of a server and is based on the request frequency characteristic of a user and the corresponding relationship characteristic for analysis and statistics. Through the determination whether the user's request behavior eigenvalues exceed a threshold value or not, it is possible to identify illegal users and malicious computer software. With a high success rate of identification, the method and system of the invention cannot be hacked easily.

The invention relates to a method and a system for identifying the bar code of a server-assisted mobile terminal, belonging to the technical field of the mobile communication value-added services. The method comprises the following steps of: collecting bar code images, resolving the bar codes, invoking the server-assisted resolving capability, transmitting through the network connection, managing the network connection, assisted-resolving the bar codes, processing resolution results, and finally showing identification results to a user at the bar code identifying client. The invention provides a two-stage identification method including the mobile terminal direct identification and the server-assisted identification. The method provides a two-stage identification method consisting of mobile terminal direct identification and server-assisted identification and comprising the following steps of: firstly, using a camera of the mobile terminal to directly scan the bar codes for identification, and judging whether to invoke the server function for assisted identification on the basis of the identification results, thereby ensuring the identification efficiency of the terminal bar codes and simultaneously achieving the goals of improving the identification success ratio and the stability. Therefore, the implement of the method is practical and feasible.

The invention discloses a ship water gauge reading automatic identification method based on multi-model fusion. The method comprises the following steps: (1) carrying out water gauge area detection onan original image; (2) cutting out a local image of the water gauge from the original image; (3) carrying out image processing on the local image of the water gauge; (4) correcting the original imageto obtain a corrected original image; (5) digging out the corrected local image of the water gauge, and performing scale character detection to obtain a scale character detection result; (6) correcting a scale character detection result to obtain a water gauge scale reading; (7) establishing a scale reading model; (8) performing image segmentation on the corrected original image to obtain waterline position information; (9) inputting the waterline position information into a scale reading model to calculate the scale reading of the waterline of the water gauge; according to the invention, theproblem of automatic reading of the current ship water gauge can be effectively solved, and the working efficiency is greatly improved; and the method can be reused in other application scenes related to water gauge reading.

The invention relates to a method for identifying the type and the characteristic of crude oil through near-infrared spectrum. The method comprises that various types of crude oil samples are collected, and the physical data is determined by using a conventional method; the absorbance at the near-infrared spectrum regions 4628-4000 cm<-1> and 6076-5556 cm<-1> and the physical data corresponding to the crude oil sample are taken to establish a crude oil sample near infrared spectrum database; the absorbance of a crude oil sample to be identified at the characteristic spectrum regions of the near-infrared spectrum forms a vector x; the x is calculated, the identification parameters Qi of each sample in the crude oil sample near infrared spectrum database are calculated through movement correlation coefficients, and if all Qi are not greater than the threshold value, a plurality of samples having the maximum Qi are selected to form an adjacent database; a virtual database is established by using the adjacent database through a uniform distribution method; and the vector x is identified in the database by using the method to obtain the virtual database sample consistent with the crude oil sample to be identified, wherein the physical properties are the physical data of the crude oil sample to be detected. According to the present invention, with the method, the existing crude oil sample physical data can be effectively utilized, and the crude oil sample identification rate can be improved.

The invention discloses a microstrip phased array antenna applied on an ETC terminal information acquisition system. The microstrip phased array antenna comprises the following three parts: a feeding network which is composed of an RF switch, a crossover and a power divider and is used for distributing signals with different powers and different phases to different ports of a Butler matrix and inputting the signal to a preset port in switching; a wave beam forming network which is composed of improved Butler matrixes and is used for generating the signal with the required phase and inputting to a preset antenna unit; an antenna array, wherein each antenna unit is composed of double radiation sheets with a 90-DEG phase difference. The antenna is mounted right above the middle line of a middle lane. The projection of the antenna in the normal line direction is in the middle line of the middle lane. The microstrip phased array antenna can settle technical problems of wave beam scanning, lane identification, identification distance increase, identification success rate increase, etc.

A speaker recognition method through emotional model synthesis based on Neighbors Preserving Principle is enclosed. The methods includes the following steps: (1) training the reference speaker's and user's speech models; (2) extracting the neutral-to-emotion transformation / mapping sets of GMM reference models; (3) extracting the emotion reference Gaussian components mapped by or corresponding to several Gaussian neutral reference Gaussian components close to the user's neutral training Gaussian component; (4) synthesizing the user's emotion training Gaussian component and then synthesizing the user's emotion training model; (5) synthesizing all user's GMM training models; (6) inputting test speech and conducting the identification. This invention extracts several reference speeches similar to the neutral training speech of a user from a speech library by employing neighbor preserving principles based on KL divergence and combines an emotion training speech of the user using the emotion reference speech in the reference speech, improving the performance of the speaker recognition system in the situation where the training speech and the test speech are mismatched, and the robustness of the speaker recognition system is increased.

The invention relates to a power distribution network high-resistance grounding fault recognition method based on a convolutional neural network. According to the method, first, a power distribution network high-resistance grounding fault and a three-phase voltage signal and a zero-sequence voltage signal at the low-voltage side of a main transformer under multiple types of transient disturbance are acquired; second, a local feature scale method is utilized to decompose the signals, equal-bandwidth band-pass filtering is performed on all the voltage signals, a time-frequency matrix is constructed, and a block time-frequency spectrum is obtained; and last, a convolutional neural network algorithm is adopted to perform classification and recognition, and whether the high-resistance groundingfault occurs is judged.

The invention discloses a method and an apparatus of displaying and recognizing images on the basis of a curved screen. The method and apparatus relate to the technical field of image processing with the object of solving problem in the prior art that when a user scans an image displayed on a curved screen, the image is recognized at a low rate. The method comprises the production of a new image based on the curving degree of the curved screen and a to-be-corrected image. The to-be-corrected image comprises a to-be-corrected two-dimensional code or a to-be-corrected bar code. On the curved screen, the new image is displayed and when the new image is mapped on a plane, the size of the new image is identical to that of the to-be-corrected image.

The invention discloses a virtual information display method and a virtual information display device, which belong to the field of Internet technologies. The virtual information display method comprises the steps of: extracting feature information of an image from the image acquired by a camera of a terminal in real time; if a feature database does not comprise feature information matched with the feature information of the image, acquiring current geographical position information of the terminal and orientation information of the camera, wherein the feature database is at least used for storing feature information and geographic position information of a plurality of targets; determining a matched target from the feature database according to the current geographical position information of the terminal and the orientation information of the camera; and acquiring and displaying virtual information of the matched target. The virtual information display method and the virtual information display device can improve the success rate of identification, especially in the case that the image comprises a plurality of targets or the target is a large-scale target such as a building or a scenic spot, and can prevent the occurrence of identification failure.

The invention discloses a health detection system adaptive to height. The system comprises a face recognition module, a body temperature detection module, a control center, an input module and a height detection module. The face recognition module and the body temperature detection module are connected to a support perpendicular to the ground; and the control center controls a driving unit to drive the input ends of the face recognition module and the body temperature detection module to point to the forehead part of the detected face. The working position is automatically adjusted according to different human body heights, and the forehead part of the human face at the fixed point position is photographed to obtain the human face recognition and body temperature data, so that the infraredtemperature measurement error can be effectively avoided, and the human face recognition success rate is improved; due to the adoption of a comprehensive detection method, the human health index datacan be comprehensively acquired in real time, so that the user experience is improved; and the system has the characteristics of high accuracy, safety and stability.

The invention discloses a power transmission line identification and positioning method based on a binocular camera and a depth camera. The invention belongs to the field of machine vision. A three-dimensional coordinate of a target is measured under a complex background; the binocular camera and the depth camera are fused; the method comprises the steps that firstly, a system is built; the systemis composed of the binocular camera and a Kinect depth camera, wherein the binocular camera collects a color image; the Kinect depth camera collects a color image and a depth image; the binocular image is matched with the Kinect image respectively; through the depth information of the depth camera, the image information outside a specified depth area in the binocular image is filtered, so that the filtering of the complex background in the binocular image is realized, and then the processed binocular image is processed, segmented and matched to obtain the three-dimensional coordinate and attitude information of a power transmission line.

The invention belongs to the technical field of a verification code, and in particular relates to a method for rapidly identifying a hollow verification code. The method comprises the steps of firstly, thinning a character contour line of the verification code in a preprocessing stage; secondly, only filling a character block through inner and outer contour lines to obtain a solid character; thirdly, segmenting the solid character to obtain single characters and character components; fourthly, analyzing structural characteristics and order associations of the character components, and combining the character components by use of a minimum nearest neighbor algorithm to obtain the non-redundant single characters; and finally, identifying the single characters obtained after segmentation andcombination by use of a convolutional neural network to obtain a final deciphering result. According to the method for rapidly identifying the hollow verification code, compared with an existing identification method, not only is an original structure kept from being not damaged, but also breakages of the character contour lines are guaranteed to be accurately repaired, the filling time can be shortened, the filling precision can be improved, accurate acquisition and rapid identification of the hollow identification code are realized, the identification success rate is greatly improved, the applicability is high and better application values are provided.

The invention provides a biological feature recognition method and a mobile terminal, and aims at solving the problem that the recognition rate is low as corresponding biological features only can berecognized according to currently started biological feature recognition manners of existing mobile terminals when the mobile terminals recognize biological features. The recognition method comprisesthe following steps of: receiving a target operation input by a user; in response to the target operation, obtaining scene feature information of environment where the mobile terminal is located; anddetermining at least one target biological feature recognition manner from a plurality of preset biological recognition manners according to a relationship between the scene feature information and recognition conditions corresponding to the plurality of preset biological feature recognition manners, and recognizing target biological features, input by the user, corresponding to the target biological feature recognition manner. According to the method, target biological features with relatively high recognition success rates are selected from a plurality of preset biological features to be recognized under the current scene feature information, so that the recognition success rate of the mobile terminal during the biological feature recognition is improved.

The invention discloses a speech control method, a speech control device, a storage medium and an air conditioner. The method includes the following steps that: a speech instruction for controlling adevice to be controlled is acquired; whether the identity of a user who issues the speech instruction belongs to a set identity range is determined; if the identity of the user belongs to the set identity range, a customary speech system matching the identity of the user is called to perform semantic analysis on the speech instruction; and if the identity of the user does not belong to the set identity range, a set default speech system is called to perform semantic analysis on the speech instruction. With the speech control method, the speech control device, the storage medium and the air conditioner of the invention adopted, the problem of the low success rate of speech recognition which is caused by failure to recognize the language of a user under a condition that the language of the user does not comply with a selected language system can be solved, and the success rate of speech recognition can be improved.