233results about How to "Solve the noise problem" patented technology

The invention discloses a method for extracting a video interested region based on a visual attention. The interested region extracted by the method combines a static image domain visual attention, amotion visual attention and a depth perception attention, and the method efficiently restrains the internal unicity and inaccuracy of each visual attention extraction and solves the problem of noisescaused by a complex background in the static image domain visual attention and the problem that the motion visual attention can not extract an interested region with small local motion and motion amplitude, thus the method improves the calculation accuracy, enhances the algorithm stability and extracts an interested region from a background with complex veins and the motion environment; furthermore, the interested region obtained by the method accords with the depth perception feature interesting on an object having a strong depth perception or a short distance in the stereovision and the semantic characteristic of human eye stereovision besides according with the vision feature of human eye interesting a static grain video frame and interesting a motion object.

The invention provides statistically optimal near-field acoustical holography used for visual recognition of air-conditioner noise sources and an operation method thereof. The operation method comprises the followings: determining the positions, frequency characteristics and distribution ranges of the noise sources, and distance between microphones on microphone arrays, number, sample rates and number of sampling points of microphones; calibrating arrayed microphones; collecting data; recording sound pressure data; processing the sound pressure data; reading the processed data and finally entering noise source recognition procedures. By the invention, complex sound pressure on reconstructive surfaces and vibration velocity of surface particles can be calculated directly via linear superposition of the complex sound pressure on a holographic plane in a spatial domain, thus essentially overcoming window effects and wraparound errors existing in near-field acoustical holography and avoiding the problems brought about by the demand of a large number of microphones by near-field acoustical holography, such as low reconstruction efficiency of sound fields and costly cost of measuring systems, etc. The invention is characterized by small number of demanded microphones, low measuring cost and high reconstruction efficiency of the sound fields.

The invention relates to a bone-conduction headphone which comprises an upper cover (1) and a lower cover (2), wherein the upper cover (1) and the lower cover (2) can form a cavity (4) for containing a bone-conduction loudspeaker assembly (3) after closed; the bone-conduction loudspeaker assembly (3) is arranged in the cavity (4); the lower cover (2) is formed by one-step injection molding and is provided with a vibration plane (21); and the bone-conduction loudspeaker assembly (3) comprises a magnet (31), a coil (32) and a buffering ring (33), wherein the magnet (31) is fastened on the inner wall (211) of the vibration plane (21) of the lower cover (2), the buffering ring (33) is sheathed on the magnet (31) and is provided with the coil (32). The invention has the advantages of almost no noise, deep bass, simple structure and low cost.

The invention discloses a shift register and a grid driving device, which are used for eliminating noise of an output terminal of the shift register and improving the working stability of the shift register. The shift register provided by the invention comprises an input module, a pull-up module, a reset module, a first pull-down control module, a second pull-down control module and a pull-down module, wherein the input module is used for supplying input signal voltage to a pull-up node, which is an output node of the input module; the pull-up module is used for storing the input signal voltage and supplying first clock signal to an output terminal; the reset module is used for supplying power source cathode voltage to the pull-up node; the first pull-down control module is used for supplying the power source cathode voltage to a first pull-down node; the second pull-down control module is used for supplying the power source cathode voltage to a second pull-down node; and the pull-down module is used for supplying the power source cathode voltage to the pull-up node and the output terminal.

A neural network based detection of noise-source in air conditioner is carried out by: extracting dynamic characteristics; training neural network; and detecting noise source in air conditioner. It has distributive and parallel processing, non-linear mapping, self-adapting and robust error tolerance so as to be widely used in pattern identification, control optimization, intellectual data processing and fault diagnosis. Noises generated from different parts of air conditioner are detected in mode of neural network to determine effectively where a main noise exists with high robustness and intelligence. It can be used for noisy faults diagnosis of air conditioner or some electric appliance so as to qualify them.

The invention discloses a method for repairing and reproducing old films based on deep learning, which includes the following steps: S1, extracting the frames of a video through ffmpeg, and forming training data sets of a de-interlacing scanning model, a video frame interpolation model, a de-blurring network and a super-resolution model respectively; S2, training a de-interlacing scanning networkmodel; S3, training a video frame interpolation network model; S4, training a de-blurring network; S5, training a super-resolution network; and S6, training a de-noising network. In the invention, oldfilms are repaired by using de-interlacing scanning, video de-noising, video de-blurring, video frame interpolation and super-resolution technologies based on deep learning. Compared with manual operation, the stability is higher, the computing speed is improved, and the accuracy of image restoration is improved. The method has the advantages of good restoring effect of processed images, high clarity of restored images, convenience of use, low cost, and the like.

The present invention discloses an image processing device and method, and belongs to the field of image processing technology. The device includes: a sliding module, used to slide in an image in Bayer format to be processed by using an M*N sliding window; a read module, used to read a pixel value of a center pixel point when the center pixel point in the sliding window is a green pixel point; a calculating module, used to calculate a symmetric value of the pixel value of the center pixel point according to the sliding window; a determining module, used to determine an equilibrium output value of the center pixel point according to the pixel value and the symmetric value of the center pixel point; and a generating module, used to generate a processed image according to the equilibrium output value of each green pixel point in the Bayer image to be processed. The problem existing in the background technology, i.e., noise exists in the image due to crosstalk between adjacent pixel points in the image in the Bayer format obtained by Bayer array filtering, is solved; and influences of the crosstalk between the adjacent pixel points in the image are reduced; and the image display effect is improved.

The invention provides a DC small current constant-current source and a method for adjusting the constant-current source and aims at solving the problem of the noise of the output current caused by the noise of an amplifier in a feedback circuit of the existing DC small current source. The DC small current constant-current source comprises a voltage source, a differential amplification circuit, a V / I conversion circuit, a current output terminal and a feedback circuit; the voltage source is connected in series with the differential amplification circuit and outputs voltage to the differential amplification circuit; the differential amplification circuit is connected in series with the V / I conversion circuit; the V / I conversion circuit generates the current and outputs the current by the current output terminal; the feedback circuit comprises a sampling circuit and a filter circuit which are connected in series; after sampling at the output current terminal of the V / I conversion circuit, the sampling circuit outputs the voltage signal which is fed back and filtrated by the filter circuit to the input terminal of the differential amplification circuit. Meanwhile, the output voltage of a feedback branch is taken as the following output voltage of the current output terminal. The invention can provide the DC small current with high precision.

The present invention provides a magnetic levitation transferring system which comprises the following components: a loading unit which is loaded with object; a body transferring unit while the loading unit is coupled to the body transferring unit at one side; a rail which forms a line for transferring the body transferring unit; a levitating unit which is connected with the body transferring unit and causes the body transferring unit to levitate to a preset height relatively to the rail; a pushing unit which is connected with the body transferring unit and provides a pushing force for transferring the body transferring unit levitated by the levitating unit along the line of the rail; and a non-contact power source which receives power through a non-contact mode and supplies power for at least one in the levitating unit and the pushing unit. Through the configuration, a magnetic levitation technique is adopted in transferring object as one of the non-contact techniques thereby settling the problems of assembly damage caused by wearing and abrasion, noise generation and particle generation. Furthermore the power is supplied stably and safely while the object is transferred with a high speed.

The invention belongs to the technical field of microelectronics, and particularly relates to a silicon-based CMOS image sensor and a method for improving electron transfer efficiency of the silicon-based CMOS image sensor. According to the silicon-based CMOS image sensor and the method for improving the electron transfer efficiency of the silicon-based CMOS image sensor, due to the fact that increasing doping from the interior to the surface is led into the portion where a light sensing area makes contact with a transfer transistor (TX) channel area, the electric potential is gradually increased from the interior of a substrate to the TX surface channel, photon-generated carriers can be completely transferred, the theoretical limit of the number of the transferred carriers can be met, and the problem of noise caused by incomplete transfer of the photon-generated carriers is successfully solved.

The invention discloses a transmitting end, a receiving end, a system and a method for power line communication of a contact network and belongs to the technical fields of railway communication and urban rail communication. The transmitting end comprises the following modules including an application layer encoder, a physical layer encoder and a physical layer modulator, namely an OFDM (Orthogonal Frequency Division Multiplexing) modulator, wherein the application layer encoder is used for receiving source information and providing an encoding data stream; the application layer encoder comprises a CRC (Cyclic Redundancy Check) encoding and fountain code coding cascade; the physical layer modulator is used for receiving a data stream of the application layer encoder and has functions of forward error correcting and encoding; and the physical layer modulator is used for receiving a data stream of the physical layer encoder and generating an output signal by using an OFDM modulation mode. The invention further discloses a corresponding receiving end, a system and transmitting and receiving methods. Due to the combination of a fountain code cross-layer technology, a physical layer channel encoding and OFDM modulation technology, the problem of five kinds of noises of general power line communication is solved; and particularly due to the adoption of the fountain code cross-layer design, the problem that the on-off of a contact network channel and a current collector are intervened by electric spark pulse noise of the contact network is solved.

The invention relates to a method for identifying the grabbing stability of a manipulator based on deep learning, and belongs to the technical field of robot perception. The method in the invention comprises the steps of: acquiring tactile data in different grabbing stabilities at first, then, converting time series picture data into a picture, training a deep learning network by utilizing tactile picture data, and finally, identifying the tactile picture data having the unknown stability by utilizing the deep learning network. By means of the method in the invention, training of the deep learning network is carried out based on original data different in time duration, so that the obtained network has the robustness in data duration. The deep learning network in the method is used for updating network parameters based on integrated input data; and influence on the network is very low only when labels of a few data have errors. Thereby, the method has the robustness to noise data.

The invention discloses a partial discharge positioning device and method for a power equipment inspection unmanned aerial vehicle. The device comprises a partial discharge detection device which is installed on the unmanned aerial vehicle and comprises an antenna, an electromagnetic signal conditioning circuit, an ultrasonic sensor, an ultrasonic conditioning circuit, an analog-to-digital conversion circuit, a microprocessor, an airborne WiFi communication module and a power supply; and a ground portable computer which is used for receiving the detection result information processed by the microprocessor for storage and display and can control the partial discharge detection device. The device can quickly evaluate the insulation state of the 10 kV overhead insulated wire, solve the problem that the unmanned aerial vehicle of the 10 kV overhead insulated line cannot determine the partial discharge position information quickly in the course of inspection, improve the anti-noise abilityof the system and avoid the use of expensive ultrasonic arrays and antenna arrays. The device also has the advantages of small size and low weight and is easy to install on the unmanned aerial vehicleand has great engineering practical value.

The invention provides a robot, a map construction method, a positioning method, electronic equipment and a storage medium. The map construction method of the robot includes the steps that the robot traverses a working region according to a preset rule, and an initial map is constructed according to images which are collected in real time when the robot traverses the working region, and comprisesa first map body and a second map body, wherein the first map body includes mapping of the working region and a map coordinate system, and the second map body includes extracted scene features based on the collected images and a geometric quantity of the scene features. The map is constructed without active signals by the adoption of a passive vision sensor, the vision sensor is low in power consumption and cost, the quantity of obtained signals is large, and the map construction of the intelligent robot is optimized.

The invention discloses a design method of a variable-structure controller for double teleoperation with the feature of time-varying delay. Conventional impedance control is adopted by a master terminal. Variable structure control is adopted by a slave terminal. When a variable-structure controller at the slave terminal is designed, the method of integrating errors is adopted in order to decrease a noise problem caused by differential terms and serves as a sliding mode function. The design method of the variable-structure controller for double teleoperation with the feature of time-varying delay has following beneficial effects: a sliding mode controller is reasonably selected in order to reduce negative impact of time delay to a great extent; meanwhile, the sliding mode controller is not required to determine concrete value of time delay so that the noise problem caused by derivation of the sliding mode function is avoided; furthermore, fine robustness and stability are obtained without influence of changes and time delay; and the requirement for a double teleoperation system featuring time-varying delay can be satisfied.

The invention relates to an MP3 compressed domain audio self-adaptation noise reduction method. The method is directly used for noise reduction on MP3 compressed domain. Firstly, MDCT coefficient is extracted from MP3 audio data containing noise, and activity detection is carried out on MP3 audio based on MDCT spectrum energy characteristic, so as to distinguish active audio band and mute band. Meanwhile a normal inverse Gaussian (NIG) distribution function is adopted to carry out prior statistic modelling on the MDCT coefficient according to the sparse property of the MDCT coefficient after MDCT coefficient is extracted form MP3 compressed audio data. Then a maximum posterior probability estimator based on NIG prior probability model is designed according to Bayesian theory, so as to obtain corresponding audio band attenuation factor. In the attenuation noise part, the attenuation factor is utilized to attenuate noise of the audio band, and attenuation iteration number is adaptively adjusted according to the attenuation weight of mute band audio, so as to realize noise reduction. Experiment result shows that noise in MP3 audio can be effectively eliminated by adopting the noise reduction algorithm of the invention, signal to noise ratio of compressed audio is improved, and MP3 audio after noise reduction has good quality.

The invention relates to a prediction method for an influence of an interior part on vehicle sound insulation and sound quality. The prediction method comprises the following steps: step 1, adjustingthe working condition of the vehicle; step 2, arranging a sensor; step 3, collecting noise in the vehicle under various working conditions; step 4, checking the consistency of data; step 5, processingand analyzing the noise data; step 6, obtaining virtual "digital interior part"; and step 7, predicting the noise level of the tested vehicle. According to the prediction method for influence of interior part on vehicle sound insulation and sound quality, the final sound insulation level of the complete prototype in the early stage of vehicle development can be predicted, the research and development costs can be reduced, and the development efficiency of the vehicle can be improved.

The invention relates to the field of intelligent power grids, in particular to a multi-station fusion system operation situation anomaly monitoring and alarm merging method, which comprises the stepsof training an anomaly detection model based on historical data of a power multi-station fusion system, and determining an abnormal value threshold; obtaining time sequence data of the operation situation of the power multi-station fusion system in real time, inputting the processed time sequence data into a trained anomaly detection model to obtain an anomaly score of the data, comparing the anomaly score with an anomaly value threshold, and determining anomaly information; based on the historical alarm data information, mining the correlation between the alarm information, and determining an association rule; and determining the category and the emergency degree of the anomaly information, generating corresponding alarm information, and completing combination and output of the alarm information based on the association rule. According to the invention, combination and output of alarm information are realized, and effective detection of anomaly data in time series data in operation and maintenance of the power multi-station fusion system is realized.

The invention discloses a face detection tracking method and device. The method comprises: acquiring a to-be-detected image and a target face image; detecting and describing feature points in the to-be-detected image and the target face image; matching feature points in the to-be-detected image and the target face image according to detection and description results; and performing face detectionand tracking according to a matching result. According to the method, the problem of accumulation during feature point extraction and the problem of noise during feature point description in an existing face detection tracking method are solved, and the face detection tracking accuracy is improved.

The invention discloses a full decoupling steering-by-wire system. The full decoupling steering-by-wire system is used for solving the problems that a motor in a traditional EPS is large in noise, andthe function of steering full decoupling needed by an intelligent automobile cannot be achieved; mechanical connection between a steering wheel and a steering wheel is omitted in parts of steering-by-wire systems, and consequently, efficacy-losing steering cannot be completed when the systems are powered off and lose efficacy; when the parts of the steering-by-wire systems adopts motors as steering-by-wire power sources, the noise is large, and the response speed is low; and when the parts of the steering-by-wire systems adopts liquid or gas as a steering medium, leakage occurs, the failure rate is increased, and maintenance is difficult. According to the full decoupling steering-by-wire system, electromagnetic force is adopted to drive a steering rod to move, so that an automobile completes a steering action rapidly and precisely; and the full decoupling steering-by-wire system can achieve full decoupling of a driver and the steering system, has the efficacy-losing protection function, and is free from constraints of the motors, the liquid and the gas, simple in structure and high in integration level.

The invention relates to a silencer drainage structure of an engine exhaust system. The silencer drainage structure comprises a silencer tube body and is characterized in that multiple rows of inner drainage holes are formed in the silencer tube body in the tube wall circumferential direction. Multiple outer drainage holes are formed in the middle of a resonant cylinder, the resonant cylinder sleeves the silencer tube body and wraps multiple rows of inner drainage holes, two ends of the vibrating cylinder are fixedly connected with the silencer tube body, a resonant cavity is formed between the outer wall of the silencer tube body and the inner wall of the resonant cylinder, and silencing cotton is filled in the resonant cavity. The silencer drainage structure is simple and reliable, can timely drain water accumulated in a silencer pipeline, prevent pipeline corrosion and can further reduce noise.

The invention is suitable for the field of mobile communication, and provides a method for automatically compensating the ascending optical loss gain of an optical fiber repeater and the near end of the optical fiber repeater. The method provided by the invention comprises the following steps: combining optical signals transmitted through optical fibers at the multi-path far end by utilizing an optical combiner at the near end; detecting the optical power of the optical signals output by the optical combiner to obtain the optical power value actually accepted; calculating the optical loss value required to be compensated according to the optical power value actually accepted, the combination number of the optical combiner and the preset lossless value, and outputting the corresponding radio frequency gain compensation value; and compensating by a gain regulator at the near end according to the radio frequency gain compensation value. By utilizing the superposition and loss characteristics of the optical combiner on each path of far-end optical power, the embodiment of the invention directly realizes the automatic compensation at the near end, solves the noise problem and does not need to consider the complicated gain allocation at the near end and the far end.

The invention relates to an audio data processing method and an audio data processing device. The method comprises the steps of acquiring adjacent first and second audio frames from an audio data stream, wherein the first audio frame is in prior to the second audio frame in time sequence; determining a frame segmentation position, wherein sampling point values at the frame segmentation positions in the first audio frame and the second audio frame meet a distance approximate condition; acquiring the sampling point value before the frame segmentation position in the second audio frame and the sampling point value after the frame segmentation position in the first audio frame, splicing the sampling point values in sequence to generate a third audio frame, and inserting the third audio frame between the first audio frame and the second audio frame. According to the audio data processing method and the audio data processing device provided by the invention, the noise generated when the audio frame is inserted can be substantially overcome.

The invention discloses a transferring system for magnetic levitation, which comprises a common support frame, a track, a suspension unit, a propulsion unit and a power supply unit.A loading unit for loading objects is coupled to the common support frame.The track is equipped with a route for transmitting the common support frame.The suspension unit comprises a plurality of sub suspension units apart from each other and connected with the common support frame, wherein the common support frame suspends to a preset height with respect to the track.The propulsion unit connected with the common support frame is used for providing propulsive force, so that the suspended common support frame is transmitted along the route of the track.The power supply unit supplies power for the suspension unit and the propulsion unit.By adopting the above configuration, when various objects (including transmission boxes), especially large-sized objects are transmitted, magnetic levitation method is adopted as one of non-contact methods, thereby avoiding the damage of components due to friction and abrasion, the noise problem and the generation of particles.Meanwhile, the suspension unit can be easily and correctly controlled.

The invention relates to an image identification technology, and provides a curvature feature recurrent neural network-based three-dimensional target identification method for the problem of image noises in a three-dimensional target identification process in order to effectively describe features of a three-dimensional target under different view angles. The method comprises the steps of firstly,obtaining a joint curvature of a target three-dimensional model by calculating a local average Gaussian curvature and an average mean curvature of the target three-dimensional model, forming a curvature sketch of the three-dimensional model by extracting a local maximum value of the joint curvature, and generating a 360-degree two-dimensional image sequence by utilizing transmission projection transform to serve as an input of a trained recurrent neural network; and secondly, obtaining an identification type with a maximum correct probability by utilizing a softmax function in a softmax layerthrough utilizing a bidirectional recurrent neural network (BRNN) as a three-dimensional model multi-view sequence feature learning method. According to the method, common features of the three-dimensional target and a two-dimensional image can be automatically extracted; and relatively good robustness and relatively high target identification rate can be kept under image noise conditions.

The invention discloses a method for measuring laser linewidth of an ultra-narrow linewidth laser. The method is creatively characterized in that by use of an automatic heterodyne detection system, a power spectrum of the ultra-narrow linewidth laser under an automatic heterodyne conduction is obtained; and a peak and a valley which are adjacent to each other in terms of positions are selected from an area beyond high-peak pulses on the power spectrum, a difference delta S between the contrast of the peak and the valley on the power spectrum are respectively calculated, and the laser linewidth delta f is calculated according to the delta S. The method for measuring the laser linewidth of the ultra-narrow linewidth laser, brought forward by the invention, has the following technical advantages that the method can realize measurement of ultra-narrow linewidth under the condition of quite short delay fibers, greatly reduces hardware demands, can effectively prevent the problem of noise brought by large-length delay fibers and is quite good in measurement precision.

The invention provides a new energy vehicle and an automatic acoustic system and an automatic acoustic control method thereof in order to solve the problems that in the prior art, only low-frequency noise or high-frequency noise needs to be eliminated, the processing mode is not comprehensive enough, and the sound quality of the sound environment still needs to be further improved. The automatic acoustic control method comprises the following steps: acquiring a noise signal in a sound environment of a new energy vehicle, generating a harmony structure signal according to the high-frequency noise signal, according to the low-frequency noise signal, generating a phase de-noising signal which is inversely equal to the low-frequency noise in amplitude, playing the harmony structure signal, playing a noise reduction structure sound fused with the high-frequency noise in the new energy vehicle sound environment so as to perform noise reduction processing on the high-frequency noise, and playing the phase de-noising signal which is inversely equal to the low-frequency noise in amplitude so as to eliminate the low-frequency noise. According to the automatic acoustic control method providedby the invention, no matter what working conditions are in the vehicle, a driver and passengers have a very high-quality sound quality environment.

The invention discloses an entity relationship extraction method based on an improved deep residual network and a relationship word attention mechanism, which comprises the following steps: convertingEnglish into a word vector by processing a remote supervision entity relationship data set; obtaining relation words, and calculating cosine similarity to extract relation word weights of sentences;respectively introducing a segmentation mechanism and a gating mechanism to form a segmented residual error network with a gate module, and segmenting the characteristic channel by adopting the segmented gating residual error network; splicing the average pooling layer on the basis of the maximum pooling layer at the rear section of the residual network to form a double-pooling layer, wherein thedouble-pooling layer is used for reserving sentence structure features; obtaining final sentence features of the model; and performing relationship classification on the obtained sentence features toobtain an output entity relationship. According to the method, the deep residual error network is applied to a remote supervision data set of relationship extraction, the noise reduction capability ofthe model is effectively improved by deepening the number of network layers, and meanwhile, a segmented gating residual error module is provided, so that the performance of the deep residual error network is effectively improved.

The invention discloses a ventilation louver with adjustable resistance and noise reduction, which includes a louver frame and noise-eliminating blades. The ventilation louver also includes a sensor and a control terminal. The sensor is arranged on one side of the louver frame to detect temperature and / or changes in wind power, and the sensor is electrically connected to the control terminal, and the control terminal is electrically connected to the sound-absorbing blade. The invention can satisfy the air inlet pressure loss and noise reduction performance of the equipment, and simultaneously has the effects of achieving low carbon, environmental protection and energy saving.