96results about How to "Realize the recognition function" patented technology

The invention provides a voice recognition system comprising a storage unit, a voice acquisition and pre-processing unit, a feature extraction unit and a pattern matching unit, wherein the storage unit is used for storing at least one user's voice model; the voice acquisition and pre-processing unit is used for acquiring a to-be-identified voice signal, transforming the format of the to-be-identified voice signal and encoding the to-be-identified voice signal; the feature extraction unit is used for extracting voice feature parameters from the encoded to-be-identified voice signal; and the pattern matching unit is used for matching the extracted voice feature parameters with at least one voice model so that the user to which the to-be-identified voice signal belongs is identified. Based on the voice generation principle, voice features are analyzed, MFCC (mel frequency cepstrum coefficient) parameters are used, a speaker voice feature model is established, and speaker feature recognition algorithms are achieved, and thus, the purpose to improve speaker detection reliability can be achieved, and the speaker identification function can be finally realized on an electronic product.

The invention discloses an OLED display panel and a display device. An image sensor is arranged below OLED devices; and light shading layers, each of which has at least one small hole imaging region,are arranged between the image sensor and the OLED devices. With small hole imaging regions corresponding to gaps among all OLED devices and being staggered with shading components in signal routing units and control devices in the light shading layer, imaging of an object above the OLED display panel is carried out at the image sensor; for example, when a finger is placed above the OLED display panel corresponding to the small hole imaging region, fingerprint acquisition and recognition are realized. Therefore, the fingerprint recognition function is realized in the display area of the OLED display panel; and thus the attached area of the whole display device can be reduced and a frameless design can be realized.

The invention provides an image object recognition method based on SURF (Speed Up Robust Feature), comprising the following steps: first, preprocessing images; second, extracting SURF corners and SURF descriptors of the images to describe the features of the images; third, processing the features through PCA data whitening and dimension reduction; establishing a bag-of-visual-words model through Kmeans clustering based on the features after processing, and using the bag-of-visual-words model to construct a visual vocabulary histogram of the images; and finally, carrying out training by a nonlinear support vector machine (SVM) classification method, and classifying the images to different categories. After classification model building of different images is completed in the training phase, the images tested in a concentrated way are detected in the testing phase, and therefore, different image objects can be recognized. The method has excellent performance in the aspects of recognition rate and speed, and can reflect the content of images more objectively and accurately. In addition, the classification result of an SVM classifier is optimized, and the error rate of judgment of the classifier and the limitation of the categories of training samples are reduced.

The invention discloses a method for training a speaker recognition network model. The method comprises the steps of acquiring the ternary group of i utterance from the training data set, and inputting the ternary group of the i utterance into a convolutional neural network, the convolutional neural network carrying out characteristic extraction on the ternary group of the i utteranceto obtain a ternary group of the i utterance characteristic and inputting into the linear neural network; inputting the first identity vector information of the first speaker and the second identity vector information of the second speaker into a linear neural network for fusion processing, so as to obtain the fused ternary group of the i utterance; and calculating a ternary group loss according to the ternary group of the i utterance so as to adjust the network model. The speaker recognition network model obtained by the embodiment of the invention can realize the recognition function of speakers more accurately, and has the lowest error rate.

The invention discloses an automatic identification and security early warning method for a vehicle novice driver. The method comprises the following steps: firstly, analyzing operating characteristics of the vehicle novice driver on a vehicle steering wheel, an accelerator pedal and a brake pedal on the basis of mass data regarding to the condition that the novice driver operates a car at the urban road environment, and building a novice driver driving behavior ethogram; and acquiring information of the vehicle steering wheel rotation angle, the accelerator pedal stroke and the brake pedal stroke by using a vehicle-mounted CAN bus, judging whether the current vehicle driver is a novice driver through analyzing the operating characteristics of the current vehicle driver on the steering wheel rotation and the accelerator/brake pedal on the basis of the comparison between the driving data of the vehicle driver and features of the novice driver driving behavior ethogram by adopting the dynamic time warping algorithm, and setting a more reasonable safe distance and early warning time on a vehicle safety assistant system for the novice driver.

The invention relates to a fast lane line detection method based on parameter space voting and Bessel fitting. The method comprises the steps of obtaining an image in front of a vehicle, converting the image into a grey-scale map; firstly, extracting feature edge points based on line scanning; extracting lane feature information through fast inverse perspective transformation to remove interference edge points; putting the edge points into a self-defined parameter space, voting, counting and screening the edge points belonging to the lane, performing Bezier curve fitting on the screened laneedge points to generate a curve cluster, calculating the distance between the curve cluster and the edge points, and selecting a Bezier curve fitting lane closest to the edge points from the curve cluster and the edge points. According to the technical scheme, the method based on the lane line vanishing points is adopted to automatically demarcate the region of interest, remove redundant information and increase the detection speed, edge feature points are subjected to fast inverse perspective transformation, lane features are more remarkable, interference points can be eliminated, and the real-time performance of the algorithm and the reliability of the lane lines are improved.

The invention relates to the technical field of display, and discloses a display panel, a preparation method thereof and a display device. The display panel comprises a substrate, a thin film transistor array and a photosensitive diode array located on the substrate in sequence, and also comprises a first passivation layer, a first planarization layer and a second passivation layer located betweenthe thin film transistor array and the photosensitive diode array and sequentially arranged in the direction from the thin film transistor array to the photosensitive diode array. Via holes are formed in the first passivation layer, the first planarization layer and the second passivation layer, and the photosensitive diode array is electrically connected with the thin film transistor array through the via holes. According to the display panel, the photosensitive diode array structure layer is integrated, the fingerprint induction and recognition functions of the display area can be achieved,the dark-state current of the photosensitive diode structure is low, the contrast ratio of fingerprint imaging is high, and imaging is clear.

The invention discloses a false fingerprint recognition method. The method includes the following steps: (1), arranging a fingerprint collection end; (2) arranging a recognition storage module connected with the fingerprint information collection end; (3), embedding a fingerprint image information processing algorithm in the recognition storage module; (4), acquiring a to-be-recognized fingerprint image A through a camera; (5), acquiring a to-be-recognized fingerprint image B through a pressure-sensitive sensor; (6), matching the to-be-recognized fingerprint image A with the to-be-recognized fingerprint image B, and if the to-be-recognized fingerprint image B cannot be generated or the to-be-recognized fingerprint image B can be generated clearly while the to-be-recognized fingerprint image A and the to-be-recognized fingerprint image B deviate in fingerprint position, indicating that fingerprint is false; if the to-be-recognized fingerprint image B can be generated clearly and the to-be-recognized fingerprint image A and the to-be-recognized fingerprint image B are completely coincident in fingerprint position, indicating that the fingerprint is true, and completing false fingerprint recognition. The invention further discloses a device implementing the method.

The invention discloses a multi-sensor based smart automobile environment perception method. The technical scheme aims to solve the technical problems of single perception measure and difficulty in accurately, comprehensively and fully extracting information of an environment surrounding an automobile in an existing automobile environment perception system. According to the technical scheme, detection of objects surrounding the automobile by 360 degrees and a front lane line recognition function are realized by a multi-perception environment sensor consisting of an ultrasonic radar sensor, a millimeter radar sensor and a camera sensor through a sensor data preprocessing algorithm, a multi-sensor data fusing algorithm and a target tracking algorithm, and 360-degree perception for the automobile can be realized truly.

The invention discloses an intelligent aroma volatilizing system which comprises a sensing device, a control center, a volatilizer, a client and a cloud, wherein the sensing device is used for collecting and extracting environmental parameter information influencing volatilization and volatilizing state in an external environment, converting the environmental parameter information into an electric signal, and transmitting the electric signal to the control center; the control center is used for controlling the volatilizer by combining historical control information of the cloud and the behavioral habit of the user according to an input instruction of the user at the client and the volatilization environmental information transmitted by the sensing device, and transmitting the operating state of the system to the client; the volatilizer is used for accepting a control signal of the control center, and controlling multidimensional volatilization of aroma. The intelligent aroma volatilizing system disclosed by the invention is controlled by a mobile terminal to finally produce high-end intelligent home equipment which can be used for simulating multiple scenes; the high-end intelligent home equipment has a main function of simulating smell, and realizes sense similarity of environmental atmosphere and ingredient similarity of healthy health-care functions.

The invention relates to a fluorescence sensing material for Cu2+ and Fe3+ detection and a preparation method and application thereof. The preparation method includes dissolving polyphenylene vinylene precursor and polymethyl methacrylate into N, N-dimethyl-formamide solvent, obtaining polyphenylene vinylene precursor/polymethyl methacrylate nanofiber membrane after electrostatic spinning, and subjecting the obtained nanofiber membrane to a heat eliminating treatment under a vacuum state at the temperature of 110 to 130 DEG C to obtain the fluorescent sensing material. The fluorescent sensing material has a high specific surface area, so that the contact surface of the fluorescent sensing material and detection molecule can be effectively enlarged to obtain greatly improved detection sensitivity. The fluorescent sensing material has no specific receptor and has a special porous nanofiber structure formed by high temperature elimination to realize identification of the substance to be detected. When subjected to optical excitation, the fluorescent sensing material shows high selectivity and sensitivity to Cu2+ and Fe3+ and has different responsiveness. The fluorescent sensing material can be reused and has a good application prospect.

The invention discloses a heavy-fog scene aerial image fusion recognition method based on a self-adaptive cloud model. The method comprises the following steps of acquiring an original heavy fog sceneaerial image as a sample, and performing graying processing on the heavy fog scene aerial image by using a weighted graying algorithm, and obtaining a grayscale image; processing the grayscale imageby using histogram equalization and obtaining a balanced grayscale image, carrying out edge detection on the balanced grayscale image to obtain an edge image, and initializing a convolution neural network and training data, and extracting feature data; performing multi-scale decomposition on an image subjected to training through a heavy-fog scene aerial image recognition training model and obtaining a smooth image and a detail image, and the image is reconstructed to obtain a fused image, and the feature data extracted by inputting the fused image into the convolutional neural network model are transmitted to a support vector machine for training, and the extracted test characteristic data are input into a large fog scene aerial image recognition training model for judgment, and finally the accurate large-fog scene aerial image recognition result is obtained.

The embodiment of the invention provides an article identification recognition method, device and system applicable to multiple applications. The article identification recognition method applicable to the multiple applications comprises the following steps: when article identification information is received, encoding characteristic information and data information of an article identification are analyzed from the article identification information; according to the encoding characteristic information, a matching decoding operation is conducted on the data information so as to acquire recognition information of the article identification information; the recognition information of the article identification information is sent. The article identification recognition method, device and system applicable to the multiple applications can be compatible with recognition functions of article identifications of various encoding systems, thereby solving the problem that compatible recognition of the article identifications can not be realized as a result of different encoding systems.

The invention relates to a method and device for digitally collecting and indentifying anti-counterfeiting spectroscopic analysis information of a bill. The method comprises the following steps of: respectively irradiating a bill with anti-counterfeiting ink, anti-counterfeiting material or a special color pattern by using ultraviolet lights, infrared rays or visible lights; collecting an image formed by laser lights of the anti-counterfeiting ink and the anti-counterfeiting material or reflected lights of the special color pattern; and identifying the image by referring to a standard bill image and judging the authenticity of the bill. Through the method and device provided by the invention, high-quality and fast collection of the bill image can be realized.

The invention provides a rotating unfolding disc typed multi-layer storage device. The rotating unfolding disc typed multi-layer storage device comprises a storage disc position control device and at least one layer of storage disc. The storage disc position control device comprises a base, and a motor installed on the base. Each layer of storage disc comprises a chassis, at least two sector disks, connecting rods equal to the sector discs in number, a pole plate and a driving device. The driving device is fixed on the chassis, the pole plate is connected to the output shaft of the driving device, one end of each connecting rod is connected with the pole plate in a revolute pair mode, the connecting point of each sector disc with another end of a connecting rod of the chassis also forms a revolute pair, the chassis of the lowest layer storage disc is connected with the motor output shaft of the storage disc position control device, and the chassis of the upper storage disc is connected with the lower storage disc through the protruding pin column at the bottom of the upper storage disc. The rotating unfolding disc typed multi-layer storage device is reasonable in structure design, convenient to operate, safe and reliable, and capable of recognizing surgical instruments in the work process of surgical nurse assistant robots.

The invention discloses an identification system based on a communication beam orbital angular momentum mode of a digital micro-mirror, and relates to an optical communication technology. The identification system is as follows: a He-Ne laser is communicated with the front and back of a first beam splitter; the first beam splitter is respectively communicated with a first spatial light modulator and a second spatial light modulator; the first spatial light modulator and the second spatial light modulator are respectively communicated with a second beam splitter; and the second beam splitter, afirst digital micro-mirror device, a second digital micro-mirror device, a third digital micro-mirror device, a lens and a charge coupled camera are communicated in turn. According to the identification system based on the communication beam orbital angular momentum mode of the digital micro-mirror, separation and identification of a multiplexed multi-mode orbital angular momentum beam are realized using the digital micro-mirror device; distortion correction of each light spot phase is realized utilizing a phase compensation mechanism; and mode recognition of orbital angular momentum under any mode is realized, and simultaneous mode recognition of the multi-mode orbital angular momentum beam is overall realized utilizing a diffraction identification mechanism.

The present invention provides a tumble pre-judgment method based on human body key point behavior recognition and LSTM. According to the tumble pre-judgment method based on human body key point behavior recognition and LSTM, the human body is further divided into a head area, a trunk area and a leg area for behavior recognition, the calculation amount is greatly reduced, and therefore the detection efficiency is improved; and on the basis, the memory function of the acquired video is realized by adopting an LSTM (Long Short Term Memory), namely a long-term and short-term memory neural networkmechanism, so that the functions of analyzing and identifying the behavior change of the human body are realized, and finally, the identification results are classified into three types, namely tumble, non-tumble and others. According to the method, the calculation power consumption is reduced, and the fall detection time is saved, so that the functions of real-time detection and fall detection pre-judgment are realized.

The invention belongs to the field of information management, and provides an online encoding system of electronic supervision codes. The system is on the same production line, an encoding portion is utilized for encoding a product, under the condition that the production line does not stall, the quality checking portion is utilized for performing online recognition on the electronic supervision codes, the encoding function and the recognition function on the electronic supervision codes can be achieved on the same production line at the same time through cooperative work between the encoding portion and the quality checking portion, and therefore the encoding efficiency of packaging box skin is improved, the encoding cost is reduced, and the encoding technology is optimized.

The invention relates to an electronic artificial hand based on sound control, which includes an artificial hand speech recognition module, a signal amplification output chip and a signal relay control circuit. And the speech recognition module is composed of a speech recognition chip which can only recognize the user's own speech signal; and the speech recognition chip input end receives at least 7 kinds of speech instructions, one speech instruction is used for specialized training of urgency stop controls of any actions of artificial hand, other six instruction are outputted for the artificial hand electric relay control circuit, two to six instruction output signal are respectively used for controlling one to three freedom degrees of artificial hand, thereby finishing six actions of finger extension, finger bending, wrist internal rotation, wrist outward rotation, wrist extension and wrist bending. The invention can realize speech study function and recognition function, and realize outputs of two to more than six control instructions. The electric relay is used for controlling positive and negative rotation of electric motor, thereby realizing opening and closing functions of artificial hand controlled by two signals.

The invention discloses an insulator evaluation and classification method and device based on infrared images. The method comprises the steps of acquiring an insulator infrared spectrum sample library; cutting the obtained images of the insulator infrared spectrum sample library to obtain a normal insulator library as a positive sample and a deteriorated insulator library as a negative sample; graying the images in the positive sample and the negative sample; processing the image after graying processing by using a principal component analysis method; and training the image processed by the principal component analysis method and the characteristics of the insulator state by adopting a Bayesian algorithm to obtain a corresponding relationship between characteristic parameters and the insulator state and model prediction parameters, thereby realizing classification and identification of the insulator. According to the technical scheme of the invention, the method can adapt to a complexfield environment, achieves the automatic recognition and diagnosis of the insulator in the infrared image, effectively reduces the false detection, and improves the efficiency and accuracy.