711results about How to "Recognition speed is fast" patented technology

A pattern recognition system and method are provided. Aspects of the invention are particularly useful in combination with multi-state Hidden Markov Models. Pattern recognition is effected by processing Hidden Markov Model Blocks. This block-processing allows the processor to perform more operations upon data while such data is in cache memory. By so increasing cache locality, aspects of the invention provide significantly improved pattern recognition speed.

A method for automatically recognizing one or more structures in digitized image date includes determining sub-jets of the image graph at the nodes defined by its structure. The sub-jets correspond to at least part of the determined sub-jets of the specific reference graph. The projection of the net-like structure of the specific reference varies until a graph comparison function, which compares the jets of the image graph with the corresponding jets of the specific graph, becomes optimal. The structure is associated with a reference image corresponding to the reference graph for which the graph comparison function is optimal, with respect to the optimal image graph determined for the reference graph.

Disclosed is a method to improve iris recognition success ratio and speed by registering the iris code on location of illuminator. In the present invention, iris codes for a single iris generated as many as the number of a plurality of illuminators are registered on an iris database every area corresponding to the location of the illuminator. The iris codes generated one by one whenever the plurality of illuminators are turned on one by one for recognition are searched on the corresponding areas of the iris database to determine the recognition.

The subject invention leverages a scalable character glyph hash table to provide an efficient means to identify print characters where the character glyphs are identical over independent presentation. The hash table allows for quick determinations of glyph meta data as, for example, a pre-filter to traditional OCR techniques. The hash table can be trained for a particular environment, user, language, character set (e.g., alphabet), document type, and / or specific document and the like. This permits substantial flexibility and increases in speed in identifying unknown glyphs. The hash table itself can be composed of single or multiple tables that have a specific optimization purpose. In one instance of the subject invention, traditional OCR techniques can be utilized to update the hash tables as needed based on glyph frequency. This keeps the hash tables from growing by limiting updates that reduce its performance, while adding frequently determined glyphs to increase the pre-filter performance.

The invention discloses a two-dimensional image recognition and object reconstruction method based on a three-dimensional model database. The method includes the following steps: a three-dimensional model of an object used for a two-dimensional projection is constructed, the three-dimensional model database is established and an correspondingly correlative attribute database is built up; the two-dimensional projection matching degrees of a two-dimensional image of the object to be recognized and the three-dimensional model of the relative object are compared; as for the two-dimensional projections being in accordance with the matching degree requirement, the object in the correlative attribute database is recognized as the object in the two-dimensional image; and the three-dimensional model in the three-dimensional model database which is correlative to the object in the attribute database is used for reconstructing the object. Due to the adoption of the comparison of the two-dimensional image of the object with the two-dimensional projection of the relative three-dimensional model, the aim of recognizing the object in the two-dimensional image of the object is realized and the two-dimensional images of the object which are photographed in a plurality of angles and from a plurality of directions are recognized; moreover, the aim of reconstructing the object in the two-dimensional image is achieved by using the correlative three-dimensional model; in addition, a paralleling calculation method is adopted to accelerate the speed of the reorganization and the reconstruction.

The present invention puts forward a front vehicle information structured output method base on concatenated convolutional neural networks, for mainly solving the problem that a traditional method cannot quickly detect and identify a vehicle body, a license plate and a vehicle logo in one time in a complex environment. The realization process of the front vehicle information structured output method comprises the steps of: 1, acquiring a sample set and selecting a vehicle body initial sample set from the sample set; 2, training the vehicle body initial sample set through a BING (Binarized Normed Gradients) method; 3, respectively training convolutional neural networks of vehicle body, license plate and vehicle logo; 4, judging the area and color of the vehicle body according to the well trained convolutional neural network of vehicle body; 5, judging the area of the license plate and identifying a license plate number according to the well trained convolutional neural network of license plate; 6, judging the area and class of the vehicle logo according to the well trained convolutional neural network of vehicle logo; and 7, outputting the all obtained information to a frame image. The front vehicle information structured output method of the present invention can accurately detect and identify front vehicle information in a complex environment, and can be used for intelligent monitoring, intelligent traffic, driver auxiliary and traffic information detection.

The invention provides a device and method integrating gesture recognition and ultrasonic tactile feedback. The device integrates a functional unit comprising a camera, an illumination unit and an ultrasonic spray head assembly inside gloves, feeds back the hand operation in real time, achieves accurate identification and tracing to the hand actions of a user and converts tactile information fed back into forces or force moments directly acting the hands of the user, so that an operator can produce a real touch feeling in a virtual environment, and meanwhile touch feeling at multiple angle and strengths can be also simulated, the gesture recognition speed and accuracy are improved, and the fidelity and immersion degree of virtual reality experience are increased. The invention also relates to application of the device integrating gesture recognition and ultrasonic tactile feedback in virtual reality, augmented reality and mixed reality.

The present invention discloses a facial emotion recognition method based on a depth sparse self-encoding network. The method comprises the steps of 1, acquiring and pre-processing data; 2, establishing a depth sparse self-encoding network; 3, automatically encoding / decoding the depth sparse self-encoding network; 4, training a Softmax classifier; and 5, finely adjusting the overall weight of the network. According to the technical scheme of the invention, sparseness parameters are introduced. In this way, the number of neuronal nodes is reduced, and the compressed representation of data can be learned. Meanwhile, the training and recognizing speed is improved effectively. Moreover, the weight of the network is finely adjusted based on the back-propagation algorithm and the gradient descent method, so that the global optimization is realized. The local extremum and gradient diffusion problem during the training process can be overcome, so that the recognition performance is improved.

The invention relates to the technical field of PMSMs, solves a coupling problem during online identification of multiple parameters of a surface-mounted type PMSM, and achieves online decoupling identification of PMSM inductance, stator resistance and rotor flux linkage. Accordingly, the technical scheme adopted by the invention is that an online decoupling identification method of multiple parameters of the PMSM comprises the steps as follows: 1) identifying and coupling analysis of parameters of the PMSM; 2) a decoupling identification strategy, wherein voltage deviation before and after D shaft current injection is used for increasing the order of a motor mathematical equation, so that the decoupling identification of multiple parameters of the surface-mounted type PMSM inductance, the stator resistance and the rotor flux linkage are achieved; 3) neural network identifier design, wherein according to a parameter online identification problem of the PMSM, online identification is performed on motor parameters by adopting a self-adaptive neural network structure and a weight convergence algorithm based on a least mean square algorithm. The method is mainly applied to the design and manufacture of the PMSM.

The invention discloses an ultrasonic phased array inspection oil-gas pipe circular weld defect automatic recognition method, in particular to an ultrasonic phased array inspection oil-gas pipe circular weld defect automatic recognition method for a support vector machine, which belongs to a pipe circular weld defect type recognition technology. The recognition method includes two parts of the construction of a support vector machine model and the automatic recognition of unknown defect types in circular weld. The construction steps of the support vector machine model include artificial defect signals extraction, wavelet decomposition promotion, feature extraction, feature optimization and the training of the support vector machine model. The automatic recognition steps of unknown defect types in circular weld include defect signals extraction, optimized feature calculation and automatic recognition based on the support vector machine. The invention combines a lifting wavelet transform with a fractal technology and has the advantages of high recognition accuracy and fast recognition speed. The automatic recognition of the defect types can be realized.

The invention provides an automatic expression recognition method. The method includes the following steps: step10, capturing face areas in frame images of a video and preprocessing the face areas, step20, extracting Gabor features and EOH features of images of the corresponding face areas, step30, integrating the corresponding features to acquire final superficial features of the target video and acquiring an expression tag sequence of each frame of video images through a classifier acquired by training, step 40, scanning the expression tag sequences, judging duration time of expressions and outputting expression classes according to the acquired micro- expressions.

The dynamic programming technique employs a lexical tree that is encoded in computer memory as a flat representation in which the nodes of each generation occupy contiguous memory locations. The traversal algorithm employs a set of traversal rules whereby nodes of a given generation are processed before the parent nodes of that generation. The deepest child generation is processed first and traversal among nodes of each generation proceeds in the same topological direction.

The invention discloses a regional convolutional neural network-based method for gesture identification and interaction under egocentric vision. The method comprises the following steps of S1, obtaining training data; S2, designing a regional neural network which is used for gesture classification and fingertip detection while being used for hand detection, and ensuring that an input of the neural network is a three-channel RGB image and outputs of the neural network are top left corner coordinates and top right corner coordinates of an external connection matrix of a gesture region, gesture types and gesture skeleton key points; and S3, judging the gesture types, and outputting corresponding interactive results according to different interactive demands. The invention provides a complete method for gesture identification and interaction under egocentric vision. Through single model training and partial network sharing, the identification speed and accuracy of the gesture identification under the egocentric vision are increased and improved.

The invention provides a table recognition method and system, belonging to the technical field of table recognition, which utilizes nonlinear contrast enhancement based on weighted RC threshold iteration and Gaussian Laplacian LoG operator to carry out binarization processing on table images conforming to the format, and utilizes tilt correction algorithm based on perspective change to carry out tilt correction; uses an image morphological processing method to extract table box lines, segmentation of cells, to obtain the smallest cell; establishes the character database of the smallest cell, carries on the neural network training, establishes the form recognition model, carries on the recognition to the form. The invention has the advantages of simple calculation, fast speed, accurate identification of table images with weak contrast, uneven light and dark distribution of images and blurred background. The speed and precision of recognition are improved by establishing special high-frequency characters, training special neural network and template matching. At the same time, the customized neural network is simple in structure and reduces the time and workload of training and tuning.

A prescription verification system includes a database that contains a plurality of spectral signatures corresponding to identified pharmaceuticals. A multimodal multiplex sampling (MMS) spectrometer obtains a spectra of a pharmaceutical to be identified and verified. The pharmaceutical can be inside or out of a vial. The prescription verification system includes algorithms for matching spectra of pharmaceuticals to be verified obtaining using the MMS spectrometer to spectral signatures contained in the database corresponding to identified pharmaceuticals. The prescription verification system further includes algorithms for identifying such pharmaceuticals to be verified.

The invention relates to a method for the contactless detection and recognition of gestures in a three-dimensional movement space (3D gestures), whereinin an electrical near-field which defines the three-dimensional movement space, deformations of the electrical near-field are detected which can be performed by movements of at least one object, e.g. one or two fingers in the three-dimensional movement space,at least one movement path is generated from the detected deformations of the electrical near field, which corresponds to the movement of the at least one object in the three-dimensional movement space, andwherein during the generation of the movement path in the movement path a gesture start is determined, wherein beginning with the gesture start the gesture is extracted from the movement path.The invention also relates to a system for the contactless detection and recognition of gestures in a three-dimensional movement space.

The invention discloses a method for controlling a transport robot to autonomously enter an elevator. The method for controlling the transport robot to autonomously enter the elevator comprises the following steps of navigating the robot to a doorway of the elevator by utilizing an own navigation device of the robot, before the robot enters the elevator, obtaining the position of an elevator call button by utilizing a robot vehicular vision module with depth information collection, triggering the elevator call button according to the position information of the elevator call button, and calling the elevator; obtaining the state information of an elevator door by utilizing a retreating sensor module, and controlling the robot to enter the elevator. According to the method for controlling the transport robot to autonomously enter the elevator, through combining with signals obtained by a Kinect sensor and the retreating sensor module, the past way that an instruction is sent to an elevator controller by a remote control center is thoroughly changed; the past way is changed into the way that an instruction is sent to the robot by the remote control center, the instruction is sent to a PLC (Programmable Logic Controller) by a robot vehicular controller and the elevator is triggered through an arm of the robot; by combining with the redundant design of a full load judgment condition module in the elevator, the control safety performance of the elevator is promoted.

The invention provides a speech recognition modeling method and a speech recognition modeling device. The above speech recognition modeling method comprises steps: speech signals are converted into a feature vector sequence and an annotated text corresponding to the speech signals is converted into a modeling unit sequence, wherein each modeling unit in the modeling unit sequence is a complete initial consonant or vowel producing unit or an initial consonant or vowel producing unit related to a context; a blank label is added in front of or behind any modeling unit in the modeling unit sequence; and based on connectionist temporal classification (CTC), the feature vector sequence and the modeling unit sequence after the blank label is added are trained, and a speech recognition model is built. According to the CTC-based deep neural network initial consonant and vowel modeling of the invention, the recognition speed and the recognition accuracy for the built speech recognition model can be improved.

The invention relates to a device and a method for on-line three-dimensional measuring of a laser radar of heavy forging in the measuring technical field, which comprises a two- dimensional laser radar, a servo-motor, a junction box, a rotary main shaft, a spacing and null point sensor, a bearing block which is provided with a bearing, a band brake apparatus, a back supporting stand of the laser radar, an L-shaped mounting bottom plate, a vertical mounting chassis, a movement-control card, a data collecting card, and a computer which is used for processing data. Firstly, the laser radar is positioned on the horizontal surface of a forging axial cord, then four surfaces of the forging are respectively scanned by the laser radar, the two-dimensional laser scanning radar is scanned in the surface which is vertical to an axial cord of the forging, the servo-motor drives the laser radar to rotate in the horizontal surface, thereby the three-dimensional scan of the forging is realized, finally, shape and parameter of the forging are gained through data process, and deviation of the shape of the forging is gained. The invention greatly increases accuracy and speed of identification, enlarges range of application, satisfies measuring requirement of a forging, and effectively increases measuring accuracy and efficiency of the forging.

The invention relates to an airport intelligent luggage car or a high-speed railway station luggage car, which belongs to the field of airport, high-speed railway station, shopping mall conveying, orfiled of unmanned conveying of materials in hospitals or factories. The conventional system cannot deal with the situation that flow density is quite large in dynamic environment. Compared with the conventional system, the invention has the following innovations in aspects that multisensor fusion identification of pedestrians and moving targets and location and localization and mapping: 1) a global map is formed through the collaboration of each car, and updating and refining are conducted based on the local environment information uploaded by each car in real time; 2) multiple sensors, a laser radar, vision, a mile meter, an inertial measuring unit (IMU)) are fused, greatly improving positioning precision; 3) the vision and the laser radar are fused, rapidly and effectively identifying pedestrians and moving targets; 4) automatic charging and self-returning are conducted, saving a large amount of labor; 5) the car can be parked in a parking area in an inserting manner, saving parkingspace and achieving good order. The whole system is strong in practicality and easy to be implemented.

The invention discloses a license plate character recognition method based on the combination of K-L transformation and LS-SVM. Firstly, the method of K-L transformation is adopted to carry out dimensionality reduction of features to a character image of the license plate; then, according to the permutation features of the character image of the license plate, the cluster distance method of cluster analysis is adopted, and four groups of optimal LS-SVM classifiers of binary trees are designed to respectively realize the recognition of English alphabets, numbers, characters and English alphabets plus numbers in license plate characters. The method adopted by the invention better solves the problem that other license plate character recognition methods have unidentifiable fields, does not require to traverse all classifiers when classifying, thus greatly improving classifying efficiency, shortening the feature extraction process of the license plate characters, reducing the calculation work of sample training and the recognition time of the license plate characters and simultaneously improving the recognition rate of a license plate recognition system and the capability of classification and promotion.

The present invention relates to a handwriting processing method and system, characterized in that the method comprises: collecting handwriting input by a user from an input area, and storing the handwriting as a handwriting picture; when all the handwriting is saved as a handwriting picture, Recognizing and storing all handwriting pictures as editable text in a unified manner; displaying the handwriting picture and / or the editable text according to user selection. In the present invention, after all the handwriting is saved as a handwriting picture, the step of identifying the handwriting picture as a text is unified, and the text can be edited, viewed, retrieved, shared, and copied, which not only greatly improves the input efficiency, but also facilitates users to choose according to their needs. Images and / or editable text can be flexibly switched between displays.

The invention proposes a spatio-temporal data stream video behavior recognition method based on deep learning. Compared with a traditional deep learning method which utilizes a signal data stream, namely a single video stream, the spatio-temporal data stream video behavior recognition method based on deep learning utilizes spatio-temporal data streams, that is a spatial stream and a temporal stream, the temporal stream recognizes category of a target in a video from a static video frame, the spatial stream recognizes motion of the target from motion components in the video, and classification results of the temporal stream and the spatial stream are integrated to obtain a final behavior category. The spatio-temporal data stream video behavior recognition method based on deep learning recognizes the target and the motion separately, can reduce calculation burden of a neural network, and effectively increases accuracy.

The present invention relates to a method of applying a pattern recognition method to a digital spectrometer to carry out nuclide rapid identification. The method comprises the steps of using the digital spectrometer to carry out real-time acquisition and data pre-processing on a nuclear radiation pulse signal, and selecting the characteristic energy ray and the area increase ratio of the characteristic energy peak of a correlated nuclide as the needed characteristic information, wherein the characteristic energy ray can be used as the main characteristic to identify the existence of the nuclide, and the area increase ratio of the characteristic energy peak is used as the auxiliary characteristic to be used as the criterion of the parallel analysis of a multi-ray nuclide. The advantages of the present invention are that: the area increase ratio of the characteristic energy peak of the nuclide substitutes the operation in the conventional method of selecting the full-energy peak area of the characteristic ray of the nuclide as the basis of identifying the existence of the nuclide, so that the identification of the nuclide is not interfered by the full-energy peak counting area of the characteristic ray of the nuclide accumulated in the early stage, and the identification reliability of the nuclide is improved; the characteristic extraction process maps an input pattern from the object space to the characteristic space, so that the quantity of information is compressed, the identification and judgement are easy, and the identification speed and efficiency of the nuclide are improved.

The invention provides a vehicle-mounted man-machine voice interaction system. The vehicle-mounted man-machine voice interaction system comprises a voice collection module, a voice purification module connected with the output end of the voice collection module, a voice processing method selection module connected with the output end of the voice purification module, a network module connected between the output end of the voice processing method selection module and the input end of a cloud server, a system terminal processor connected with the output end of the voice processing method selection module, the cloud server and a vehicle-mounted control unit module. The vehicle-mounted man-machine voice interaction system preferably selects the cloud server to carry out voice identification processing under the condition that wireless network signals are good in state in order to improve the accuracy and speed of voice identification, and can also start the system terminal processor when the wireless network signals are not good in order to keep normal work of the vehicle-mounted system.

The invention belongs to the technical field of medical image identification and discloses an intelligent canceration cell identification system and a method thereof, a cloud platform, a server and acomputer, which comprise an image acquisition module, a front-end processing module, an expert cloud platform, a diagnosis and suggestion module and a display module. The image acquisition module is used for acquiring the image information of a to-be-identified cell specimen. The front-end processing module is used for carrying out the compression processing on the acquired image data of the to-be-identified cell specimen. The expert cloud platform is used for building a software platform on a cloud server side to establish an image analysis system, establishing a cell image training databaseby utilizing a deep learning convolutional neural network, identifying the acquired image information to find out determined lesion cells, and outputting the determined lesion cells to the display module. The diagnosis and suggestion module is used for displaying a symptom result, uploading the result to the expert cloud platform and providing the condition identification reference information with a certain maximum probability. The display module is used for displaying the symptom result which is accurately identified. According to the invention, a lot of manual time is saved and the missed diagnosis rate is further reduced. Meanwhile, the labor cost is reduced.

The invention provides a face recognition method and device. The method comprises the steps as follows: a to-be-recognized image containing a face is acquired; the to-be-recognized image is subjected to skin color segmentation on the basis of preset color space, and an area with the face skin color to be analyzed is determined in the to-be-recognized image; the area with the face skin color to be analyzed is subjected to face detection, and a face area is located; and the face area is subjected to face recognition, and a face recognition result is obtained. According to the method, the face recognition speed can be increased, and the face recognition reliability is improved.

The invention provides an effective modeling and identification method of moving object behaviors, which comprises the following steps: step 1, firstly extracting local features and then extracting moving features of the moving object behaviors by using a feature extraction module; step 2, fusing the extracted local features and the moving features by using a feature fusion module via a subspace learning method, and reducing dimensions on feature space by using the feature fusion module; and step 3, by combining a prototype learning algorithm with a measure learning algorithm, identifying themoving object behaviors by using a behavior identification module, and evaluating the identified moving object behaviors by using the behavior identification module. In the effective modeling and identification method, the local features and the moving features are subject to feature fusion treatment, thus reducing the dimensionality of the feature space, and improving robustness of feature expression. The effective modeling and identification method has the advantages of low storage and low computational complexity, and has preferable classification and identification performance. The effective modeling and identification method can be used for intelligently monitoring abnormal behaviors in public security fields, thus an alarm is given once the abnormal behavior is found.

The invention discloses an isolated word speech recognition method based on an HRSF (Half Raised Sine Function) and an improved DTW (Dynamic Time Warping) algorithm. The isolated word speech recognition method comprises the following steps that (1), a received analog voice signal is preprocessed; preprocessing comprises pre-filtering, sampling, quantification, pre-emphasis, windowing, short-time energy analysis, short-time average zero crossing rate analysis and end-point detection; (2), a power spectrum X(n) of a frame signal is obtained by FFT (Fast Fourier Transform) and is converted into a power spectrum under a Mel frequency; an MFCC (Mel Frequency Cepstrum Coefficient) parameter is calculated; the calculated MFCC parameter is subjected to HRSF cepstrum raising after a first order difference and a second order difference are calculated; and (3), the improved DTW algorithm is adopted to match test templates with reference templates; and the reference template with the maximum matching score serves as an identification result. According to the isolated word speech recognition method, the identification of a single Chinese character is achieved through the improved DTW algorithm, and the identification rate and the identification speed of the single Chinese character are increased.