Patents
Literature
Hiro is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Hiro

171 results about "Machine recognition" patented technology

Post-processing system and method for correcting machine recognized text

A method of post-processing character data from an optical character recognition (OCR) engine and apparatus to perform the method. This exemplary method includes segmenting the character data into a set of initial words. The set of initial words is word level processed to determine at least one candidate word corresponding to each initial word. The set of initial words is segmented into a set of sentences. Each sentence in the set of sentences includes a plurality of initial words and candidate words corresponding to the initial words. A sentence is selected from the set of sentences. The selected sentence is word disambiguity processed to determine a plurality of final words. A final word is selected from the at least one candidate word corresponding to a matching initial word. The plurality of final words is then assembled as post-processed OCR data.
Owner:PANASONIC CORP

Method and equipment for providing picture verification code based on verification security level

The invention aims to provide a method and equipment for providing a picture verification code based on a verification security level. The method comprises the following steps: enabling the verification code providing equipment to acquire a verification request submitted by a user in a visited page through user equipment; according to user related information of the user, determining a verification security level corresponding to the verification request; according to the verification security level, acquiring a picture verification code corresponding to the verification request; and providing the picture verification code to the visited page. Compared with the prior art, the invention can provide picture verification codes of different levels based on verification security levels, thereby increasing the machine recognition difficulty, and simultaneously facilitating user verification and improving the use experience of the user.
Owner:BEIJING BAIDU NETCOM SCI & TECH CO LTD

Wide-field extended-depth doubly telecentric catadioptric optical system for digital imaging

Focused imaging of constant size and resolution over a wide range of distances and a wide field of view in object space is provided by a doubly telecentric catadioptric optical system including an external limiting aperture at the juncture of the focal planes of two objectives, with a large-diameter concave spherical or aspheric mirror as the primary objective, and a camera lens as the secondary objective. Constant resolution avoids rescaling of images made at different depths for machine recognition tasks such as OCR, saving computation time and cost, and increasing through-put and accuracy. For digital linescan cameras, constant resolution of the image avoids scanning objects at different line rates for different depths of objects being scanned, thus maximizing speed of the objects and, hence, throughput. The field of view is constant over the entire range of depth and is not limited by the diameter of the camera lens elements. Undesirable cropping of the image of objects close to the imaging system is prevented. The effective f / number of the system is constant over the entire range of focus, and the image irradiance is more uniform over the entire field of view than with a traditional non-telecentric camera lens.
Owner:LOCKHEED MARTIN CORP

Signal injection coupling into the human vocal tract for robust audible and inaudible voice recognition

InactiveUS6487531B1Enhancing and replacing natural excitationEasy to identifySpeech recognitionFrequency spectrumVocal tract
A means and method are provided for enhancing or replacing the natural excitation of the human vocal tract by artificial excitation means, wherein the artificially created acoustics present additional spectral, temporal, or phase data useful for (1) enhancing the machine recognition robustness of audible speech or (2) enabling more robust machine-recognition of relatively inaudible mouthed or whispered speech. The artificial excitation (a) may be arranged to be audible or inaudible, (b) may be designed to be non-interfering with another user's similar means, (c) may be used in one or both of a vocal content-enhancement mode or a complimentary vocal tract-probing mode, and / or (d) may be used for the recognition of audible or inaudible continuous speech or isolated spoken commands.
Owner:TOSAYA CAROL A +1

Verification-security-level-based audio verification code provision method and equipment

The invention aims to provide a verification-security-level-based audio verification code provision method and verification-security-level-based audio verification code provision equipment. The method comprises the following steps of: acquiring a verification request which is submitted by user equipment in an accessed page from a user by using the verification code provision equipment; determining a verification security level corresponding to the verification request according to the user related information of the user; acquiring an audio verification code corresponding to the verification request according to the verification security level; and providing the audio verification code to the accessed page. Compared with the prior art, the invention provides the audio verification codes of different levels according to the verification security levels to bring convenience to the verification of the user and improve user experiences at the same time of increasing difficulties in machine recognition.
Owner:BEIJING BAIDU NETCOM SCI & TECH CO LTD

Novel deep convolution neural network moving vehicle detection method

ActiveCN106407931AEffectively differentiate vehicle typesSmall energy functionCharacter and pattern recognitionMobile vehicleNerve network
The invention relates to a moving vehicle detection method based on a novel deep convolution neural network. The method uses a monocular camera to implement a detection algorithm of a moving vehicle in the front and brings forward a moving vehicle detection framework based a novel convolution neural network. Through the novel convolution neural network, vehicle features can be obtained very accurately, and a target vehicle can be further accurately separated out, so that a machine recognition effect can be achieved, and the target vehicle can be tracked more quickly. As far as vehicle detection is concerned, the method can be adapted to a high-speed driving environment, and provides technical guarantee for implementation of intelligent assisted driving. The method solves the problem of traffic safety, improves road vehicle throughput, reduces the malignant traffic accident rate, and reduces life and property losses. From the perspective of improving social and economic benefits, the method has great meaning in reality and broad application prospect.
Owner:浙江高信技术股份有限公司

Cover device and method for electrical connector

A detachable device for covering an end of an electrical connector, comprising a non-conductive cover, sealingly attachable, and completely covering the mating end of the connector. An external end of the cover is generally planar with electrically conductive pads corresponding to the conductive terminals of the connector, and electrically connecting the pads to the terminals. The electrically conductive pads are arranged in a pattern, oriented to provide a single line-of-sight for machine recognition.
Owner:GM GLOBAL TECH OPERATIONS LLC

Natural language machine recognition method and system

The present invention discloses a natural language machine recognition method, comprising the following steps of (1) establishing a database corresponding to similar predicate calculus form meanings; (2) inputting natural language information; (3) segmenting sentences of the natural language information one by one, and converting the sentences into one or more than one similar predicate calculus form sentence according to a segmentation rule; (4) converting the plurality of similar predicate calculus form sentences into electric signals for machine recognition, inputting the electric signals to a central processing unit to perform logic reasoning or metaphor or association creative thinking function treatment by at least one mode of search or recognition or recursion or substitution, and generating a new numeric code combination; and (5) retroactively converting the numeric code combination into a new natural sentence corresponding to the input natural language information as an output or as a learning result for storage.
Owner:王立山

Text direction correction method and device, computer device and storage medium

The invention relates to a text direction correction method and device, a computer device and a storage medium. The method comprises the steps of obtaining a to-be-recognized text image; inputting theto-be-recognized text image into a text direction positioning detection model based on deep learning to obtain coordinate information corresponding to each text positioning box in the to-be-recognized text image; determining the direction of the to-be-recognized text image according to the coordinate information corresponding to each text positioning box; and if the direction of the to-be-recognized text image is not the standard direction, converting the coordinate information corresponding to each text positioning box in the to-be-recognized text image according to the standard direction toobtain the text image converted into the standard direction and the coordinate information corresponding to each converted text positioning box. By adopting the method, the problem of machine error recognition caused by the direction error of the text image to be recognized can be avoided, and the accuracy of machine recognition is improved.
Owner:上海眼控科技股份有限公司

Machine recognition method and device between imaging device and imaging box

The invention discloses a machine recognition method between an imaging device and an imaging box. The method comprises the following steps of: reading the standard information of the imaging device; comparing the read standard information with the information stored in an imaging box chip; when the read standard information is different from the information stored in the imaging box chip, rewriting the information stored in the imaging box chip by using a controller according to the read standard information; and updating the state of the imaging box chip into a standby state and waiting for receiving a task of the imaging device. In the machine recognition process between the imaging device and the imaging box, the standard information of a printer is read by the imaging box chip and the information stored in the imaging box chip is rewritten according to the standard information so as to make the imaging box chip applied to the imaging devices requiring different matching information, therefore, the method and the device enlarge the application range of the chip, enhance the universality of the chip, accelerate the machine recognition process, change the conventional chip development process, shorten the chip development period and reduce the development cost.
Owner:APEX MICROELECTRONICS CO LTD

Machine character recognition verification

Data on a document is recognized using at least two machine recognition processes. Data from one machine recognition process is used as reference data; data formed using the other recognition process is used as verification data. If the verification data matches the reference data, machine recognition may be verified. If the verification data does not exactly match the reference data, an assessment is made of the likelihood that the verification data is sufficiently close to the reference data to suggest an error in the verification data. This may be done by applying a fitness function to the verification data, to assess the likelihood that the verification data represents a mis-recognized version of the reference data. In one embodiment, the verification data is OCR data, and the reference data is MICR data.
Owner:SYMCOR

Human-machine recognition method of CAPTCHA

The invention discloses a human-machine recognition method of CAPTCHA, comprising: collecting a sample data set including one or multiple groups of training sample data and corresponding tags set forthe groups of training sample data, wherein the tags represent attributes of users; using the sample data set to train a machine learning model; collecting real-time user data; predicting the real-time user data according to the machine learning model so as to determine a user's attribute.
Owner:ZHONGAN INFORMATION TECH SERVICES CO LTD

Human-machine recognition method, device and system

The invention discloses a human-machine recognition method, device and system. The system includes a terminal and a server; the terminal is used for collecting user voiceprint information and behaviordata information; and the server is connected with the terminal and is used for recognizing the user voiceprint information and behavior data information uploaded by the terminal according to a pre-stored model, judging whether an operation object to which the user voiceprint information and behavior data information belongs is the same as a target object in the pre-stored model, and outputting acorresponding judgment result according to the type of the target object, wherein the judgment result may indicate that the operation object is a natural person or a robot. With the human-machine recognition method, device and system provided by the invention adopted, the technical problem of low accuracy of human-machine recognition due to defects existing in human-computer recognition in the prior art can be solved.
Owner:ALIBABA GRP HLDG LTD

Generation method of picture verification code based on characters drawing

The invention relates to a generation method of a picture verification code based on characters drawing. The generation method comprises the following steps: defining a character library; generating a background color picture of the verification code, and establishing coordinates on the background color picture; selecting four to six characters from the character library at random, so as to be used as verification code characters; generating sizes and positions of the verification code characters one by one, and drawing the verification code characters and noise; adding global noise. By using the verification code generated by the generation method, the machine recognition difficulty can be increased, additionally, normal recognition of eyes is guaranteed, and the safety performance of the verification code and good user experience are improved.
Owner:ZHEJIANG FORESTRY UNIVERSITY

State machine recognition and optimization

State machines are identified from a netlist of circuit elements of a user design. Strongly connected components in the netlist are identified as candidates for analysis. The registers of each strongly connected component are identified. An optimal set of inputs and potential state transition logic is identified for the registers in the component. A set of reachable states from an initial state of the registers of a component is determined by simulating state transitions in response to permutations of input values. State machine information is created to assist compilation software in optimizing the user design. Optimizations can include identifying redundant circuit elements based on the set of reachable states and reencoding the state machine with a different state encoding scheme to reduce the amount of state transition and output logic. A subset of the set of reachable states representing a one-hot encoded state machine may be further isolated and optimized.
Owner:ALTERA CORP

Human-machine recognition method and equipment

The invention discloses a human-machine recognition method and equipment. The human-machine recognition method comprises the steps of obtaining behavior data of a client side under a running condition; if the behavior data is matched with behavior features of a corresponding human in a preset behavior analysis model, determining that the operator of the client side is a human; if the behavior data is not matched with behavior data of a human in a preset behavior data judgement model, determining that the operator of the client side is a machine. In this way, whether a human or a machine operates the client side is judged according to the judgement whether or not the behavior data of the client side in running, namely rotation data, stress data and bearing data of the client side in running, is matched with the behavior features of the corresponding human in the preset behavior analysis model, so that whether the operator is a human or a machine is recognized on the basis of the behavior features of the human operating the client side, and recognition is simple, precise and quick.
Owner:SECBOOT INFORMATION TECH CO LTD

Image processing system and image processing method

An image processing system performing image processing on input image data includes an accepting part that accepts selection of a machine recognition processing mode intended for outputting a print image on the assumption that the image, after being output, is subjected to machine recognition processing, and a processing part that performs specific image processing adapted for the machine recognition processing instead of normal image processing, when the selection of the machine recognition processing mode has been accepted by the accepting part.
Owner:FUJIFILM BUSINESS INNOVATION CORP

Man-machine recognition device and method under Web environment based on behavior difference

The invention relates to a man-machine recognition device and method under a Web environment based on behavior difference. The device is mainly used for distinguishing whether a normal user or a scanner accesses the website. The method for recognizing a scanner in the device mainly comprises the following steps: client identification, recognition method based on scanner fingerprint, a recognition method based on access frequency within short time, a recognition method based on a condition whether an access request carries Cookie information, a recognition method based on a condition whether the client can execute the JavaScript code, a recognition method based on a click event of the client. Through the comprehensive utilization of the above methods, the scanner can be effectively recognized; after recognizing the scanner, the device can prevent the scanner from further scanning the website, the security of the website is improved. The test discovers that the device has no influence to the website access and the background server, the common commercial and open source scanner at present can be recognized.
Owner:WUHAN UNIV

Atrial fibrillation recognition method based on group convolution residual error network and long/short-term memory network

The invention discloses an atrial fibrillation recognition method based on a group convolution residual error network and a long / short-term memory network and relates to the field of machine recognition of atrial fibrillation. According to the method, characteristics of electrocardiosignals of three different frequency bands are respectively extracted through three channels on a network structurebased on a group convolution residual error network and a long / short-term memory network, characteristic analysis on a time domain is further implemented through LSTM (long short term memory), and finally electrocardiosignal segments are classified into a normal segment, an atrial fibrillation segment, a segment with large noise and a segment of other rhythms. By adopting the network model, atrialfibrillation recognition accuracy can be improved in a situation of noise interference, the analysis time can be shortened, and the instantaneity of an algorithm can be improved; and on the basis ofthe group convolution residual error network, a classification accurate rate can be increased on premise that parameter complexity is not improved, benefits of a topology structure of group convolution blocks in a residual error module are taken into play, and meanwhile, data amounts of hyper-parameters can be also increased.
Owner:GUANGDONG UNIV OF TECH

Wide-field extended-depth doubly telecentric catadioptric optical system for digital imaging

InactiveUS6980249B2Constant magnification and resolutionImprove performanceTelevision system detailsPhotometryCamera lensDigital imaging
Focused imaging of constant size and resolution over a wide range of distances and a wide field of view in object space is provided by a doubly telecentric catadioptric optical system including an external limiting aperture at the juncture of the focal planes of two objectives, with a large-diameter concave spherical or aspheric mirror as the primary objective, and a camera lens as the secondary objective. Constant resolution avoids rescaling of images made at different depths for machine recognition tasks such as OCR, saving computation time and cost, and increasing through-put and accuracy. For digital linescan cameras, constant resolution of the image avoids scanning objects at different line rates for different depths of objects being scanned, thus maximizing speed of the objects and, hence, throughput. The field of view is constant over the entire range of depth and is not limited by the diameter of the camera lens elements. Undesirable cropping of the image of objects close to the imaging system is prevented. The effective f / number of the system is constant over the entire range of focus, and the image irradiance is more uniform over the entire field of view than with a traditional non-telecentric camera lens.
Owner:LOCKHEED MARTIN CORP

Vulnerable main machine recognition method and device of power system

The invention discloses a vulnerable main machine recognition method of a power system. The method includes the steps of inquiring a system risk existing in the power system, taking the system risk as a top event, inquiring bottom events causing the top event by means of a fault tree query method, calculating the occurrence probability of each bottom event by means of a Bayesian network probabilistic algorithm, sequencing the bottom events according to the occurrence probabilities, and recognizing a vulnerable main machine of the power system in the bottom events according to a sequencing result. Besides, the invention further discloses a vulnerable main machine recognition device of the power system. In the hidden risk recognizing process, the Bayesian network probabilistic algorithm is introduced to calculate the occurrence probabilities of the bottom events, the bottom events are sequenced according to the occurrence probabilities, the influence degree of each bottom event in a fault tree on the top event is quantified, and therefore the vulnerable main machine in the power system can be effectively and quickly recognized.
Owner:POWER DISPATCHING CONTROL CENT OF GUANGDONG POWER GRID CO LTD

Data amplification method based on high-dimensional space transformation and machine recognition system

ActiveCN107729926AAvoid sample overlapAvoid problems such as model overfittingCharacter and pattern recognitionPositive sampleData set
The invention belongs to the technical field of image processing and machine learning, and discloses a data amplification method based on high-dimensional space transformation and a machine recognition system. The method includes the steps of converting the background sample data from the original space to the high-dimensional space, obtaining a high-dimensional space target sample distribution based on a distribution histogram of the background sample, generating high-dimensional space target sample data, using a distance function to conduct equation set transformation, transforming the amplification data from the high-dimensional space to the original space. According to the method, the distribution histogram of the negative samples is learned, and a corresponding positive sample data set is expanded, and the problem that positive and negative sample data in a machine learning model cannot be matched is solved. The classification performance is improved, and the classification accuracy of the positive samples is especially improved. Statistical analysis is carried out on the basis of the background samples to obtain distribution of the target sample data to be generated, so thata target sample is generated, and the effectiveness of the amplification data is improved. The problem of sample overlapping and model overfitting in conventional method when new target samples are synthesized on the basis of a small number of samples can be solved.
Owner:NORTHWEST UNIV(CN)

Irregular text correction method and device, computer equipment and storage medium

The invention relates to an irregular text correction method and device, computer equipment and a storage medium. The method comprises the following steps: positioning an irregular text region in a to-be-detected image; correcting each character in the positioned irregular text region based on the irregular text correction method of deep learning, and then splicing the corrected characters to obtain a target text region such that the problem of low accuracy of machine recognition of the irregular text region is avoided, the recognition accuracy is improved, and the recognition difficulty is reduced.
Owner:上海眼控科技股份有限公司

Application of deep learning for medical imaging evaluation

This disclosure generally pertains to methods and systems for processing electronic data obtained from imaging or other diagnostic and evaluative medical procedures. Certain embodiments relate to methods for the development of deep learning algorithms that perform machine recognition of specific features and conditions in imaging and other medical data. Another embodiment provides systems configured to detect and localize medical abnormalities on medical imaging scans by a deep learning algorithm.
Owner:QURE AI TECH PTE LTD

Man-machine identification method, identification system, MR intelligent glasses and application

The invention belongs to the technical field of machine identification, and discloses a man-machine identification method, an identification system, MR intelligent glasses and application. A user completes an interaction task, and man-machine identification is carried out according to examination question completion conditions of the user; according to the man-machine identification method on theMR glasses, a question in a certain holographic image form is displayed through the MR glasses, a real person makes a decision through cognition of the question, interactive operation is carried out through eye movement interaction, 6Dof handle interaction, head movement interaction and gesture interaction, a preset question is completed, the operator is determined to be a real person if the taskis completed, and the operator is a robot if the task is not completed; or another man-machine identification method capable of being added / replaced comprises the following steps: obtaining a man-machine identification feature model through training by obtaining interactive behavior data collected in the operation process of examination questions; and comparing the interactive behavior data collected in the operation process of the examination questions with the trained man-machine recognition model to obtain the real person probability and the man-machine recognition result.
Owner:陈涛

A method and a system for obtaining an online commodity patent based on machine recognition

The invention discloses a method and a system for obtaining an online commodity patent based on machine identification. The method includes automatically crawling commodity title and picture in shopping webpage by a web crawler, segmenting a commodity title, recognizing a brand name and a commodity name of the commodity as a searching element, and searching a patent document corresponding to the commodity; according to the product name, selecting a corresponding image library according to the name of the commodity, and obtains the template image with the highest similarity with the product byusing the convolution neural network matching based on the color feature from the image library, and then extracting the contour image of the commodity based on the contour feature of the template image, and matching the outline image of the product with the drawings in the retrieved patent document, and finding the corresponding patent retrieval results, so that consumers can judge the technicaladvantages and disadvantages of different businesses on the same kind of goods through the analysis of patent retrieval results. The invention is conducive to consumers' choice when the consumers areselecting multiple similar goods.
Owner:广州集创佳禾知识产权运营有限公司

Nameplate text detection method and device, computer equipment and storage medium

The invention relates to a nameplate text detection method and device, computer equipment and a storage medium. The nameplate text detection method comprises the following steps: positioning a nameplate area in a to-be-detected image, and detecting the nameplate area to obtain contours of a plurality of text regions in the nameplate area, so as to determine feature points in a plurality of contourpoints of the text area; and performing feature mapping on the feature points to obtain target vertexes of the corresponding text areas, and performing perspective correction on the text area according to the feature points and the target vertex to obtain each text area in the corrected nameplate area, so that a guarantee is provided for subsequent machine recognition of the text area in the nameplate area, and the nameplate text recognition accuracy of a machine is improved.
Owner:上海眼控科技股份有限公司

Method, device and equipment of constructing training set and method of human-machine recognition

The invention discloses a method, a device and equipment of constructing a training set on the basis of Web server logs and a computer-readable storage medium. The method includes: structurizing the Web logs according to pre-selected fields; utilizing preset time windows to segment the structurized Web logs; through independently using an association rule algorithm for the Web logs in each time window, acquiring a frequent set corresponding to the time window; and acquiring the training set through combining the frequent sets meeting a preset support degree and confidence degree. The method, the device, the equipment and the computer-readable storage medium provided by the invention can quickly and highly efficiently construct the training set with sufficiently high precision. The invention also provides a method of human-machine recognition. The method of human-machine recognition utilizes the above-mentioned training set to construct a human-machine recognition classifier, and can beused for quickly recognizing access records.
Owner:GUANGZHOU HUADUO NETWORK TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products