105results about How to "The recognition result is reliable" patented technology

Various processes are disclosed for conducting database searches by telephone. A user is initially prompted to utter N characters of a search query, and/or to enter these characters on a telephone keypad. Based on the user's entry, a speech recognition grammar is either selected or is generated dynamically for processing the user's utterance of the complete search query. In one embodiment, the speech recognition grammar is selected based on a sequence of N telephone keys selected by the user, without requiring the user to uniquely specify the N characters to which these keys correspond. In another embodiment, the speech recognition grammar is selected based solely on utterances of the N characters by the user such that the correct grammar is selected even if an utterance of one character is misidentified as an utterance of a similar sounding character. Also disclosed are methods for generating speech recognition grammars from query logs.

The invention relates to the field of traffic image processing, in particular to a bayonet vehicle image identification method based on image features. The steps of bayonet vehicle image recognition method based on the image features includes that a template dada bank is established, the data bank stores template images which are shot with different style and different color vehicles, vehicle attribute data, vehicle body color data and vehicle scale invariant feature transform (SIFT) feature data are stored corresponding to each template image, wherein the SIFT feature data are obtained through data processing; vehicle recognition is conducted to to-be-identified images, color identification is conducted to the to-be-identified images, the template images which are fitted with the vehicle body colors are selected form the template data bank according to color identification results; the to-be-identified images are extracted the SIFT features and compared with the selected template images; the template images which are matched with the to-be-identified images are obtained; the vehicle attribute data are output, wherein the vehicle attribute data correspond to the matched template images. The bayonet vehicle image identification method based on the image features combines the vehicle color identification and the recognition based on a SIFT operator principle. Color information is added in the identification process. The defect that the SIFT operator discard the color information is overcome. Identification accuracy can be improved.

The invention brings forward an identification method for an information carrier comprising multi-type identifications. The information carrier can be a product label. By use of auxiliary positioning identification and a corresponding positioning identification detection identification algorithm, auxiliary positioning identifications on the information carrier are automatically identified, such that automatic positioning of multi-type data is realized. By use of an image identification algorithm, such information as characters, bar codes, display contents of display screens, two-dimensional codes, images and the like can be simultaneously identified, and the electronic display screens can display such contents as the characters, the bar codes, the two-dimensional codes, the images and the like, such that the information carrier has abundant information, and the identification results are accurate and reliable. Compared to bar code identification, two-dimensional code identification and image identification, character identification is smaller in characters and high in character segmentation and character identification difficulty. In order to ensure the accuracy of the character identification, the identification results are verified for the character identification by use of already known character lengths.

The invention discloses a pointer instrument positioning and identifying method based on machine vision. The position of an instrument panel in an instrument cabinet image is accurately positioned andan instrument reading is automatically recognized. The method comprises the steps of (1) acquiring the instrument cabinet image; (2) carrying out segmentation processing on the instrument cabinet image to obtain an instrument panel image; (3) preprocessing the instrument panel image; (4) acquiring all edge contours in the instrument panel image by using a Canny edge detection method; (5) findingall linear contours in the obtained edge contours by using Hough transform; (6) screening out the linear contours where a pointer in the instrument panel is located through length characteristics of the linear contours; (7) calculating an average value theta of an inclination angle between the two linear contours where the pointer is located; and (8) calculating out a pointer reading according toa linear relation between the linear inclination angle and an instrument scale. By the adoption of a positioning and detecting technology, the method has the characteristics of accurate positioning and reading, good anti-noise performance and efficient and rapid image processing.

The invention relates to the field of image character recognition, and particularly relates to an image character sequence recognition system based on a recurrent neural network; the system comprises an image character input module, a convolutional neural network and a recurrent neural network classifier; the convolutional neural network extracts characteristics of a to-be-recognized character sequence input by the image character input module, and inputs to the recurrent neural network classifier; and the recurrent neural network classifier, according to sample characteristic data and output of the last moment, realizes continuous recognition of the character sequence. According to the system disclosed by the invention, the shortage that picture segmentation is carried out before OCR recognition is overcome, the earlier stage processing of the image character recognition is simplified, and a language model does not need to be constructed additionally to carry out optimization processing on a recognition result; while the recognition accuracy rate of character and word sequences is improved better, the processing efficiency of the character recognition is obviously improved; and the system has wide application prospect in the field of image character recognition.

The invention relates to a network environment-oriented electroencephalogram identification system and a network environment-oriented electroencephalogram identification method. The identification system comprises a plurality of wearable electroencephalogram measuring devices, more than one mobile network terminal and an electroencephalogram data processing device; the electroencephalogram measuring devices are used for processing measured electroencephalograms into network electroencephalogram data and transmitting the network electroencephalogram data to the electroencephalogram data processing device for processing; the electroencephalogram data processing device is used for outputting an identification result and transmitting the identification result to a mobile network terminal for displaying. The identification method comprises the steps of measuring the electroencephalograms of a tested individual by using the electroencephalogram measuring devices and processing the electroencephalograms into the network electroencephalogram data, transmitting the electroencephalogram data to the electroencephalogram data processing device by virtue of a wireless network card of the mobile network terminal, performing segment treatment on the electroencephalograms and extracting electroencephalogram characteristic information by use of the electroencephalogram data processing device, screening electroencephalogram characteristic vector training samples based on reputation computing, and determining the input sample and the degree of membership of a fuzzy support vector machine, thereby completing the identification of the tested individual and outputting the result.

The invention belongs to the image character recognition field and relates to an optical character sequence recognition method. According to the method of the invention, CNN (Convolutional Neural Network) and RNN (Recurrent Neural Network) technologies are adopted; feature extraction is performed on a whole picture containing a plurality of characters through a CNN; identical features are transmitted to an RNN so as to be subjected to repeatedly recursive use; and continuous prediction of the plurality of characters can be realized. With the method adopted, a defect that picture segmentation is required before OCR (optical character recognition) can be eliminated, the early-stage processing process of picture character recognition can be simplified, and the efficiency of character recognition can be significantly improved; and since the RNN recursively uses output data of the last round, and in model training, a language model of dependency relationships between characters and words can be obtained through learning, and therefore, a step in an OCR method, according to which a language model is required to be additionally built for post-processing after individual characters are recognized, can be avoided; and therefore, the recognition accuracy of character and word sequences can be better improved, and the processing efficiency of character recognition can be further improved.

The invention discloses an identity recognition method for smart wearable equipment. The identity recognition method mainly comprises: an acquisition step of acquiring motion data and physiological data and/or three-dimensional action motion data of a user; a creation step of analyzing and storing the acquired motion data and physiological data and/or three-dimensional action motion data in order to establish an identity recognition database; and a recognition step of comparing the motion data and physiological data and/or three-dimensional action motion data of the user wearing the smart wearable equipment with corresponding data stored in the identity recognition database, and implementing identity recognition according to a comparison result. According to the identity recognition method for the smart wearable equipment in the embodiment of the invention, a user identity is recognized according to physiological indexes, behavioral habits, personal preferences and the like, so that a recognition result is more reliable and safer. Through application of the identity recognition method to the smart wearable equipment for mobile payment, the convenience and safety in payment are also enhanced.

The present invention discloses a method for online identifying a measurement deviation fault for a traction substation instrument transformer. The method comprises the steps of: 1) modeling data selection and preprocessing: selecting data within I contiguous time points from instrument transformer valid measurement history data when a traction substrate is idle and all instrument transformer are operating properly, to form a matrix, and conducting standardized processing therefor; 2) establishing a primary element analysis model; 3) squared prediction error (SPE) calculation of the real-time sampling set of the instrument transformer; 4)online detection of the instrument transformer measurement deviation fault; (5) calculation of the SPE contribution values of the instrument transformers; and 6) identification of the instrument transformer subject to the measurement deviation fault. With the method according to the present invention, an instrument transformer subject to a measurement deviation fault can be online identified only by processing the measured valid data of all the instrument transformers in the traction substation, with no need of additionally arranging a waveform recording device. Therefore, the requirement on the hardware is low, the implementation cost is low, and the method is simple for promotion and application.

The invention discloses an axle recognition method and system for an axle beam. The method comprises the steps that two section sets containing 4-6 sections are marked in the lengthwise direction of the axle beam to form two virtual simply supported beams; axle beam time-interval responses of the two section sets are collected and measured, two isolating responses are calculated and obtained according to the time-interval responses, and the two isolating responses are used for constructing two response time-interval curves respectively; statistics is carried out on the local peak quantity on the response time-interval curves to obtain the number N of vehicle axles or axle sets. The system comprises sensors and a data processing operation center, wherein the sensors are installed on the two section sets containing 4-6 sections and used for converting the axle beam time-interval responses into electrical signals to be output, and the data processing operation center is used for calculating the isolating responses according to the time-interval responses and constructing the response time-interval curves, so that the number N of the vehicle axles or axle sets is obtained. According to the method, the axle-beam overall responses are used for recognizing the axle distance of vehicles, the vehicle recognition result is more reliable, and then the vehicle speed and the axle distance can be recognized accurately.

The invention provides a partial discharge defect type recognition method and a system based on a convolutional neural network. The recognition method comprises steps: A, partial discharge signals with a known partial discharge defect type are acquired; B, a partial discharge information library is built, wherein the partial discharge information library comprises partial discharge waveform signals with a plurality of defect types and corresponding defect types; C, a convolutional neural network for mode recognition is built; D, a waveform and a corresponding defect type in the partial discharge information library are used as input to train the convolutional neural network; E, multiple times of partial discharge signals with to-be-decided defect types are acquired; and F, the partial discharge waveform data with the to-be-decided defect types are inputted to the well-trained convolutional neural network, and a defect type with the largest ratio in the output is used as the final defect type. A complex feature extraction and data reconstruction process in the traditional recognition algorithm is avoided, the defect type can be recognized more accurately, and realization is easy.

The invention provides a method and device for processing images. The method comprises the steps of obtaining the images to be recognized, wherein the images to be recognized comprise character images; carrying out edge cutting processing on the obtained images to obtain an effective range, containing the character images, of the images to be recognized; carrying out preprocessing on the images in the effective range of the images to be recognized to obtain the preprocessed character images; using multiple recognition engines to carry out recognition on the preprocessed character images; outputting a reliable recognition result when the recognition result of the preprocessed character images and at least one of the recognition engines is determined to be reliable. The method and device for processing the images can solve the problems that when image recognition is carried out, effective information in the images cannot be obtained accurately, and the recognition function is single.

The invention discloses a microphone array and binocular camera-based speaker positioning and recognizing method. A microphone array and a TDOA method occupying relatively few resources are mainly used for carrying out rough positioning of a speaker, and then accurate positioning is carried out in combination with a binocular camera which is higher in precision but narrow in visual field, so thatthe software expenditure is greatly reduced on the basis of realizing accurate 360-degree omni-directional sound source positioning, a system can obtain a higher refresh rate or execute other tasks atidle time; and meanwhile, a weighted average method for dynamic adjustment is further used for trading off recognition results of a sound and an image, so that the recognition results finally outputby the system are more accurate and reliable.

The present invention discloses a physiological signal emotion identification method based on subjective and objective fusion of multiple classifiers. The method comprises that: a user experiences theuse of the product, and fills in the Chinese version of the PAD emotion inventory questionnaire; the heart rate and skin electrical signals of the user are collected during the product experience process, processing and feature extraction are carried out on the two kinds of objective physiological signals; the extracted heart rate features and skin electrical features are respectively trained andidentified by using SVM classifiers; and the identification results of each classifier are expressed in the form of probability of the target category, and the identification results are normalized;weight assignment is carried out on each classifier, and the particle swarm optimization algorithm is used to optimize the weights; and finally, identification results for different emotional categories are fused, and the type of emotion with the highest identification rate is taken as the final emotional state. According to the method disclosed by the present invention, the method of multi-classifier fusion is used to balance the decision results of subjective, objective and different physiological signals, so that the final identification result is more accurate and reliable.

The invention discloses an output picture acquisition system and method of a display screen of industrial control equipment. The system comprises an image acquisition unit for shooting a real-time picture of the display screen of the industrial control equipment to acquire image information; a template parameter processing unit for providing a recognition template, setting recognition template parameters and adjusting camera parameters; an image recognition unit for performing recognition on the image information according to the current recognition template, and outputting recognition resultdata; a recognition result processing unit for storing and/or uploading the recognition result data. The output picture acquisition system disclosed by the invention realizes the non-invasive data acquisition on the output picture of the display screen of the industrial control equipment without improving the existing production equipment and establishing the connection with the existing equipment; especially in the image recognition process, the image information is recognized according to the recognition template; not only is the recognition result accurate and reliable, but also is the system easy to arrange and low in cost.

The present invention discloses a method and device for identifying a malicious website, relates to the technical field of internet, and can efficiently identify the malicious website. The method of the present invention includes: first capturing user feature information input in a log-in page, then searching for a set of trusted domain names corresponding to the user feature information in a white list, finally comparing the domain name of the log-in page with all trusted domain names in the set of trusted domain names, and if the domain name of the log-in page does not belong to the set of trusted domain names, prompting the user with a warning. The present invention is mainly used in the procedure of identifying a phishing website.

The present invention discloses a low voltage district electric energy meter phase line identification method which is as follows: a concentrator and electric energy meters appoint phase line mark code and guidance code formats; the concentrator sends phase line request communication frames to the electric energy meters in different time and in different phases to request the electric energy meters to return phase line information of each electric energy meter; the electric energy meters constitute phase line reply communication frames, add phase line mark codes into guidance codes of the phase line reply communication frames, and send to the concentrator by the phase lines of the electric energy meters; and the concentrator completes the identification of the phase lines of the electric energy meters according to the identification of data content in the phase line reply communication frames and the phase line mark codes in the guidance codes. According to the method, by the low voltage carrier wave communication mode of sending and replying in different time and in different phases, the identification works of the phase lines of the electric energy meters can be automatically completed, in the manner of adding of the phase line mark codes after the guidance codes of the communication frames, dual identification of the phase lines of the electric energy meters can be completed, and real, reliable and accurate results of the identification of the phase lines of the electric energy meters can be ensured.

The invention discloses a method for automatically identifying species of common freshwater fish based on images. The method comprises the following steps: firstly, extracting a fish outline, sequentially carrying out complanation, length unification and fish head identification treatment on the obtained outline, then extracting an included angle of the head of a fish body, a vertice of the included angle, a height of a tail caudal peduncle, and a middle point of the tail caudal peduncle, further calculating a characteristic value of the included angle of the fish head and the width-length ratio of the fish, and finally, identifying four common freshwater fish of parabramis pekinensis, crucian carps, silver carps and grass carps by taking the included angle of the fish head and the width-length ratio of the fish as characteristics. The method provided by the invention is more reliable than a method of utilizing color characteristics obviously influenced by surroundings and illuminationor utilizing fish back outline characteristics obviously influenced by fins deformation to identify the fish species.

The invention discloses a trip mode identification method based on GPS trajectory data. For trip modes lack of a GPS signal and having an incomplete signal, a subway identification method based on a rule is adopted; and, for other trip modes having normal GPS signal records, a random forest classifier is adopted, and modeling and verification are performed through a combination with the GPS trajectory data. Before trip modes are identified, 4 kinds of characteristic parameters related to mode identification are screened through a characteristic selection process, and all the real trip modes can be accurately identified. The method is not limited by the GPS data and not relied on other data sources, and the method has the advantages of high universality, simple calculation and accurate and reliable identification results; and the method is suitable for trip mode identification based on the GPS trajectory data, and is in favor of large-scale promotion of trip survey based on intelligent mobile phones.

The embodiment of the invention provides a gait recognition model building method based on myoelectric signals, a recognition method and a device. The model building method comprises the following steps of collecting the myoelectric signals; reducing the noise; adding a class tag and extract the following EMG signal characteristics: slope change rate, Willison amplitude, logarithm of variance, waveform length, and characteristics DB7-MAV; calculating the DBI index and SCAT index and obtaining the comprehensive evaluation results. The comprehensive evaluation result are randomly divided into atraining sample group and a test sample group according to a predetermined proportion, and input them into the LightGBM model for training and testing. The parameters in LightGBM model are adjusted according to the error of training set and test set, and grouped, trained, tested and adjusted repeatedly until the error of model test results accords with the data model of predetermined standard, andthe corresponding relationship between human gait type and comprehensive evaluation results is stored in the data model. The embodiment of the invention can efficiently and accurately establish a data model, and the gait recognition rate is high and the recognition result is more reliable.

The invention provides an identification method and device for tight sandstone reservoir microcrack development. The identification method comprises obtaining mercury saturation of a rock sample in a target layer under different capillary pressures, respectively calculating logarithms of the capillary pressures and a logarithm of the mercury saturation, building a scatter diagram according to the logarithms of the capillary pressures and the logarithm of the mercury saturation, performing linear fitting for target data points in the scatter diagram so as to obtain a target fitting straight line, and identifying whether a target layer micro crack is developed according to a slope of the target fitting straight line. The technical scheme provided in the invention is simple and convenient, and whether the microcrack is developed can be quantificationally identified.

The invention discloses an image motion blur removing method based on an improved cyclic generative adversarial network. The method comprises the following steps: 1, constructing a non-paired blur clear data set; 2, constructing a generator network composed of an encoder, a feature converter and a decoder; 3, constructing a discriminator network for dividing the image by a receptive field; 4, constructing a joint loss function; 5, constructing two mirrored annular GAN networks to obtain a cyclic generative adversarial network model; 6, inputting a motion blurred image to be processed into the trained model in the step 5 to obtain a deblurred image; 7, carrying out two-dimensional Fourier transform on the preliminary deblurred image obtained in the step 6, and filtering out high-frequency bright spot spectrum information to obtain an accurate clear image. According to the method, a fuzzy kernel does not need to be estimated, calculation parameters are few, the deblurring speed is high, the problems of mode collapse and gradient disappearance are avoided, and the false recognition problem of frequency domain pseudo-clearness is solved.

The invention belongs to the technical field of mode recognition, and particularly relates to a human body posture recognition system based on millimeter-wave radar. The system comprises a millimeterwave radar subsystem, a data processing and tracking positioning subsystem and an artificial intelligence classification subsystem. The millimeter wave radar subsystem transmits linear frequency modulation continuous wave signals to a detection range, collects radar echo data reflected by a human body, and outputs intermediate frequency original data after frequency mixing, filtering and analog-to-digital conversion sampling. The data processing and tracking positioning subsystem extracts various characteristics, including distance, speed, acceleration, orientation characteristics and the like, capable of representing the posture of the human body from the original data, and tracks and positions the position of the human body in real time. The artificial intelligence classification subsystem filters, classifies and judges the feature information, and determines the posture of the human body by integrating classification and judgment results. The system aims at human body posture recognition and state monitoring in real life, and can be suitable for scenes such as families, offices, factory workshops, hospital wards and nursing homes.

The invention is suitable for the field of character identification and provides a method and a system for identifying spaces in a document. The method comprises the steps of acquiring width values of gaps between all adjacent characters in each basic unit by taking lines or paragraphs in the document as basic units, and obtaining an initial gap width set corresponding to each basic unit; taking the initial gap width set as an input set, processing the input set through a space threshold calculation method, and taking an obtained space threshold as a first space threshold; and judging whether the width values of the gaps between the adjacent characters in the basic unit are greater than the first space threshold or not in sequence: if the width values of the gaps between the adjacent characters in the basic unit are greater than the first space threshold, judging that the spaces exist between the adjacent characters; and if the width values of the gaps between the adjacent characters in the basic unit are not greater than the first space threshold, judging that the spaces do not exist between the adjacent characters. The fixed space width does not need to be used for judging the spaces, so that the precision of identifying the spaces in the document is improved and an identification result is more accurate and reliable.

The invention discloses an identification method of a code parameter of a BCH (broadcast channel) shortened code. The identification method comprises the following steps of: analyzing a frame structure of a message bit stream and determining a grouping starting point of the BCH shortened code; taking a code group serial number N, taking a fixed m on a GF (generator field) (2<m>), and selecting an m-order primitive polynomial; carrying out code length nd traverse in a sustaining code length range, supplementing 2<m>-1-nd 0 to the front of each code word, and finishing a non-shortened code of a BCH shortened code word; calculating a corresponding fiducial probability value P corresponding to the selected m and the code length n; for the selected significance level alpha, if P is larger than or equal to alpha, judging the m and nd as an actual coding parameter; determining the code length at the range of 0-(n-1) code root, and calculating the probability of each code root, wherein the probability of the code root is greater than 0.9 in each code group root, and a polynomial g(x) root is generated; determining m-order primitive polynomial as a foundation, and determining the generating polynomial g(x) of the BCH shortened code; and obtaining a message bit length kd of the BCH shortened code from the order number nd-kd of the generating polynomial of the BCH shortened code in combination with the code length nd.

The invention relates to a method of identifying penumbra fiber bright spots in the astronomical image sunspot, and belongs to the field of astronomical technology and image processing. The method comprises the steps of denoising a solar photosphere image I1 using band-pass filtering, to obtain an image I2; marking an umbra region for the image I2 using a morphological reconstruction method, to obtain a binary image I3 of the umbra region; processing the identified image I3 by using a region growth method to mark a penumbra region, to obtain a binary image I4 of the penumbra region; reconstructing the image I2 using the morphological reconstruction method, to obtain a reconstructed image I5; subtracting the image I2 from the image I5 to obtain a differential image I6; binarizing the image I6 to obtain a binary image I7; removing small area miscellaneous points to obtain a binary image I8; and multiplying the image I8 by the image I3 and the image I4 to remove points in the umbra region and a granule region, and obtaining an image I9. The invention is advantageous over the identification method based on high-pass filtering, and the identification result is reliable and accurate.