56results about How to "Suppress random noise" patented technology

An imaging device includes an imager, with which an optical image of an object scene is repetitively captured. A first movement detector detects, as to each of a plurality of object scene images according to a time series output from the imager, a movement of a first feature point between the object scene image and the object scene images immediately before, and a clipper performs clipping processing on each of the plurality of object scene images on the basis of the detection result. When a still image recording operation is performed, a CPU changes an exposure time of the imager in such a direction as to shorten the time, and a second movement detector detects a movement of a second feature point between the object scene image immediately after the recording operation (reference object scene image) and the three object scene images being successive thereto out of a plurality of object scene images, and an adder adds the respective three object scene images to the reference object scene image while displacing the same on the basis of the detection result.

The invention discloses an image de-spin and stabilization method based on subarea matching and an affine model, which comprises the steps that 1, a reference frame and a current frame are subjected to image histogram equalization processing; 2, four areas distributed crosswise and an image center area are selected as matching areas of motion estimation; 3, an affine transformation model is established, and a current frame motion vector is solved; and 4, a kalman filter model is established, and motion compensation of a current frame displacement vector is achieved. In Step 2, the operand is less; a speed is high; the motion estimation capacity is favorable when translation and rotation motion coexist in an image; and an interframe estimated error is less than one pixel. According to the method, the random motion quantity of a video image can be removed with high precision; scanning motion inherent in a platform or a load is kept; and the smoothness of a dynamic image sequence is improved. The method can effectively improve the video image and inhibit random noise.

The invention relates to a three-window based common-path interference detecting method and a device and belongs to the field of optical interference detection. According to the method and the device, after parallel lights pass through three windows composed of a to-be-detected object and a rectangular window and then pass through a first lens, a one-dimensional periodic grating and a second lens in turn, the parallel lights are emitted toward an image sensor; the parallel lights are collected by the image sensor and a computer connected with the image sensor, thereby obtaining an interference image containing three patterns; the intensity distributions are respectively I-D / 3, I0 and ID / 3; and the intensity distributions are introduced into a formula, thereby calculating phase distribution of the to-be-detected object. The method and the device have the characteristics of excellent stability, low system complexity and simple phase recovering algorithm.

The invention provides a display method of the digital oscillograph. The method includes: reading the sampled data point of present frame; computing the probability which the sampled data point disappears in the same pels point; setting the corresponding relation between the probability and color; the sampled data point displays the corresponding color on the pels point. Also, the invention provides a digital oscillograph. The invention can make user know the disappear probability of every component in signal and disply by the shading color. Even the discrete digital signal, it can make the signal near to nice and smooth.

The invention discloses a method for estimating the state of charge (SOC) of a lithium battery by an extended Kalman filter algorithm. The method comprises the following steps that S1, an equivalent circuit model of the lithium battery is established; S2, a current source E is replaced with the voltage of circuit (VOC) of the lithium battery, and a first state equation and a first measurement equation of a system are established according to the equivalent circuit model of the lithium battery; S3, the first state equation and the first measurement equation are analogous to an EKF algorithm toobtain a second state equation and a second measurement equation correspondingly; S4, the SOC of the lithium battery is estimated by the EKF algorithm. The method has the advantages that the EKF algorithm does not depend on the setting of an SOC initial value, the filter process can be convergent in a short time even through the set initial value has large difference with a true value, and accurate SOC estimation is achieved.

The invention discloses a data noise suppressing method based on a residual block full convolutional neural network. A training set and a test set for suppressing earthquake noise by using a deep learning method are selected from the same data set, so that the generalization performance of a model is limited. Specific to the problem of generalization, a double residual block is fused on the basisof a Unet network according to the design principle of a network structure in order to enhance the capturing performance of the network for random noise. The data noise suppressing method is established on an end-to-end coding-decoding network structure, noise-containing earthquake data is taken as input, and the substitutive characteristics of the random noise are extracted by a plurality of convolution layers and residual blocks to construct coding; and then a plurality of deconvolution layers and residual blocks are used for constructing decoding, so that the output of the network is the noise-suppressed earthquake data. Compared with the existing earthquake data denoising method, the data noise suppressing method has the advantages that the double residual block is fused to perform secondary digestion and learning on the extracted random noise characteristics, so that the substitutive characteristics of the noise are learned more fully. Thus, the data noise suppressing method has remarkable advantages on the aspect of generalization, and can effectively suppress the random noise and protect effective signals.

A hybrid filtering method called Karhunen Loeve Transform-Wavelet (KW) filtering is presented to de-noise dynamic cardiac magnetic resonance images that simultaneously takes advantage of the intrinsic spatial and temporal redundancies of real-time cardiac cine. This filtering technique combines a temporal Karhunen-Loeve transform (KLT) and spatial adaptive wavelet filtering. KW filtering has four steps. The first is applying the KLT along the temporal direction, generating a series of “eigenimages”. The second is applying Marcenko-Pastur (MP) law to identify and discard the noise-only eigenimages. The third applying a 2-D spatial wavelet filter with adaptive threshold to each eigenimage to define the wavelet filter strength for each of the eigenimages based on the noise variance and standard deviation of the signal. Lastly, the inverse KLT is applied to the filtered eigenimages to generate a new series of cine images with reduced image noise.

A system for surveying a subsurface formation includes one or more electromagnetic sensors located at or above the surface of the Earth. The sensors are configured to detect passive-source source signals and return signals that are based on seismoelectric or electroseismic conversion of the source signal in the subsurface formation. The system includes a processor communicatively coupled to the one more electromagnetic sensors. The at least one processor is configured to align and stack the passive-source source signals and the return signals and determine a property of the subsurface formation based, at least in part, on the aligned and stacked passive-source source signals and the return signals.

In order to solve the problems of imperfect surface defect detection information, low precision, low efficiency and the like, the invention provides a defect detection method based on a polarization structured light imaging technology and an improved Mask RCNN. The method comprises the following steps: firstly, combining polarization processing with structured light three-dimensional imaging to obtain a high-definition two-dimensional physical graph and three-dimensional space information of an object; performing median filtering processing on the two-dimensional physical graph; secondly, on the basis of a Mask RCNN target recognition method, adding a K-means algorithm to carry out clustering analysis on a training set, adding branches with side edge connection from top to bottom to an original FPN structure, and combining lower-layer high-resolution features and upper-layer high-resolution features to generate a new feature map; detecting an image with defects by utilizing the improved Mask RCNN network, and classifying, positioning and segmenting the defects; finally, obtaining a series of information such as the type, position, length, width, depth and area of the defect throughdata arrangement, achieving quantification of defect data, and the object surface defect detection precision and efficiency are effectively improved.

The invention relates to a seismic signal noise reduction method and a seismic signal noise reduction system based on a dual-tree complex wavelet domain. The seismic signal noise reduction method comprises the steps of: step a, establishing a noisy seismic signal model, and converting dual-tree complex wavelet domain decomposition transform on a noisy seismic signal to obtain a noisy wavelet coefficient at each scale in each direction; step b, establishing a bivariate model for rear part wavelet coefficients or imaginary part wavelet coefficients in the same direction and corresponding modes to obtain a threshold value function, and converting bivariate threshold value processing on the noisy wavelet coefficients in the dual-tree complex wavelet domain by utilizing the threshold value function to obtain an estimated value of a noise-free wavelet coefficient; step c, and reconstructing the estimated value of the noise-free wavelet coefficient by utilizing dual-tree complex wavelet inverse transformation to obtain original seismic signals after noise reduction. The seismic signal noise reduction method and the seismic signal noise reduction system take correlation between the rear part (or imaginary part) coefficients and the corresponding modes into account, establish the bivariate model for the rear part (or imaginary part) coefficients and the corresponding modes for noise reduction, effectively suppress random noise, make the overall seismic signals clear, and improve signal quality.

The invention discloses a method for calibrating wavelength of an ultraviolet spectrograph. The method includes the steps of: (1) starting the ultraviolet spectrograph and a main control computer; (2) starting a standard light source, and preheating for a set time; (3) performing ultraviolet spectral signal acquisition and processing by the main control computer: (3.1) obtaining dark noise of a linear array CCD detector of the ultraviolet spectrograph; (3.2) collecting the ultraviolet spectral signals, and deducting the dark noise obtained in the Step (3.1); (3.3) utilizing a narrow window difference method to perform characteristic spectrum positioning; (3.4) replacing different types of standard light sources; repeating the Step (2) and Steps (3.1) to (3.3) to obtain a peak value position of a characteristic spectrum peak; (3.5) according to the peak value position obtained in the Step (3.4), utilizing a sectional fitting method to perform spectral curve fitting; (3.6) performing reconstruction on the fitted spectral curve; and completing the whole calibration process. The method for calibrating the wavelength of the ultraviolet spectrograph is easy and convenient to operate, has a strong anti-interference capability, and is high in calibration precision.

The invention discloses a total generalized variation-based infrared image multi-sensor super-resolution reconstruction method. The total generalized variation-based infrared image multi-sensor super-resolution reconstruction method mainly comprises the steps of projecting a low-resolution infrared image into the coordinate space of a high-resolution visible image, obtaining a sparse infrared image and solving a data item weighting coefficient according to the sparse infrared image; performing normalization processing on the sparse infrared image and obtaining a normalization infrared image; solving the marginal information of the high-resolution visible image through a phase equalization algorithm; constructing a data item by the data item weighting coefficient and the normalization infrared image; weighting a TGV regular term improved by a first-order gradient operator through the marginal information of the visible image and constructing a regular bound term; adding the data item and the regular bound term to construct an objective function, solving the objective function in an iterative mode through a primal-dual optimization algorithm with the normalization infrared image serving as an initial value and obtaining a reconstructed high-resolution infrared image. Experiments show that the quality of the image reconstructed by the total generalized variation-based infrared image multi-sensor super-resolution reconstruction method is high and the image is close to an original high-resolution infrared image.

A solid-state infrared imager comprises a matrix array of infrared sensing pixels which are formed as an imaging area on a semiconductor substrate and each of which contains a pn-junction thermoelectric converter element to sense incident infrared radiation, row selection lines each connected to the pixels of a corresponding row, signal lines each connected to the pixels of a corresponding column, a row selection circuit which selects and drives one of the row selection lines, and a signal readout circuit which reads out signal currents output to the signal lines from the pixels corresponding to the row selection line driven by the row selection circuit. Particularly, the signal readout circuit includes a signal line potential stabilizer which stabilizes the potential of the signal line to a constant level, and a current-voltage converter which converts the signal current flowing in a signal line to a signal voltage.

A method (50) and apparatus (40) for reducing noise in a video signal calculates (53) a discrimination value of a filter based on a magnitude difference between a current frame and a prior frame of a sequence of frames and modifies (54) the current frame using the modified discrimination value. The output of the filter is calculated as follows:g(i, j, k)=f(i, j, k)±βwherein:f(i, j, k) is the current frame input to the filter;g(i, j, k) is the current frame output from the filter; andβ is the discrimination value of the filter, which is calculated as follows:β={κδ′+χ,δ′=max{δ,δ0},δ<δ10,δ≥δ1in which:κ and X are constants less than one;δ is a magnitude difference between the current frame and the prior frame; andδ0 and δ1 are constants that satisfy δ0≦δ1.

The invention relates to an adaptive time-frequency peak filtering and de-noising method based on improved Hough transform, which belongs to the field of geophysical technology. A multi-layer fractional Fourier Hough approach is presented for linear event seismic record. Event envelope identification is carried out on seismic data. A filter trajectory equation similar to the trend of event is built. An optimal trajectory of time-frequency peak filtering is extracted adaptively. Thus, filter bias is fully reduced. The detection result can be obtained quickly under the condition of strong noise. The method has high accuracy and strong anti-interference ability.

The invention discloses a method for achieving linear structured light three-dimensional measurement by means of phase calculation in the three-dimensional sensor technology. The method comprises the steps that linear scanning is carried out on an object to be measured through projection equipment, a linear structured light image sequence is recorded through a photographic device, a fundamental frequency phase and a plurality of high-frequency cut-off phases are worked out by the adoption of a phase calculation formula of the phase measurement profilometry, an absolute phase is obtained according to a time phase unfolding method, an area with the low signal to noise ratio in a phase diagram is filtered out according to brightness modulation, and finally three-dimensional coordinates on the surface of the object to be measured are worked out according to phase and system calibration parameters. The method can be used for the linear structured light three-dimensional measurement technology comprising laser ray scanning. The method has the advantages of being high in calculation efficiency, anti-jamming capability and measurement precision.

The invention discloses a voiceprint open-set identification method with unknown category internal division capability. The voiceprint open-set identification method has relatively high accuracy and relatively good applicability. The method is realized through the following calculation scheme: taking voice data of different speakers as input of a voiceprint open set recognition system, calculating Mel-frequency cepstrum coefficient characteristics of the voice data one by one, and training an audio coding module GE2E based on a time sequence by adopting a generalized end-to-end loss function, thereby effectively eliminating ambiguity between the speakers; through audio data coding output by GE2E, in combination with a multivariate Gaussian model and a training probability step model CGDL, judging whether any audio data belongs to a known category or not, and classifying the audio data judged to be the known category; for the audio data which is judged by the CGDL as an unknown category, constructing and training an unbounded staggered state neural network, and clustering the audio data online, wherein the obtained clustering number is the number of speakers, and all the audio data in a certain cluster belong to the same speaker.

The invention discloses a casting blank surface temperature field measurement sensor and a casting blank surface temperature field measurement method, wherein the casting blank surface temperature field measurement sensor and the casting blank surface temperature field measurement method relate to the technical field of sensors. The sensor and the method mainly aim to overcome a defect of temperature field measurement inaccuracy caused by casting blank surface emissivity uncertainty, smoke, steam and oxidation scale interference, thereby realizing accurate, stable and continuous measurement for the casting blank surface temperature field. The casting blank surface temperature field measurement sensor comprises the components of an optical lens, a stepping motor, an optical modulation turntable on which a light blocking plate and an attenuation plate are embedded, a three-spectrum thermal imaging unit, and a signal acquisition processor. The three-spectrum thermal imaging unit comprises the components of two spectroscopes, two narrowband optical plates with light transmittances which are different from wavelength, and three monochromatic surface arrays CCD.

The invention discloses a deep learning mode recognition method for distributed optical fiber pipeline intrusion detection, and the method comprises the steps: carrying out the wavelet threshold denoising of an original intrusion signal, and carrying out the multi-resolution decomposition through mallat; and mapping the denoised signal into a two-dimensional image through a GAF algorithm, and then reducing the size of the image to meet the requirements of a network model. And the network model is optimized, an Adam optimizer is utilized to optimize the learning rate, a Swsh activation function is utilized to enhance the model performance, and high-speed and high-precision identification of the intrusion event is realized. The GAF facilitates CNN recognition of intrusion events with fine feature differences, and especially has a good anti-interference effect for distributed optical fiber surrounding environment factors. As the GAF does not need to carry out iterative operation, the intrusion identification speed is greatly improved. Meanwhile, the GAF algorithm is insensitive to power fluctuation in an optical path, and the robustness and practicability of the system are effectively improved.

Embodiments of the present disclosure provide a parking space detection apparatus and method, medium and electronic equipment, in which detection is performed based on motion statuses of images of parking spaces, thereby ensuring stability of the detection result; detection is performed by combining a contour method and a machine learning method, and clearing processing is performed before using the machine learning method when a scenario is unclear, to efficiently use advantages of the contour method and the machine learning method, with the processing speed being relatively fast and being applicable to various scenarios, and the detection precision is relatively high; and furthermore, performing the detection based on the stable image may suppress random noises, and improve the detection precision.

The invention discloses a method for inhibiting power supply frequency disturbance, wherein the method includes phase position tracking, noise acquisition and noise removal. Specific to the shortcoming of signal distortion easily caused by a common filter noise reduction method, the method for inhibiting power supply frequency disturbance by removing noise is provided according to the property that the noise caused by the power supply frequency disturbance is in synchronism with the power supply frequency and the noise waveform is unchanged under a certain condition, thus the process of inhibiting noise cannot cause the distortion of a real signal. The basic principle of the method is to collect noise when the signal output is zero or constant value, and average the noise periodically so as to inhibit the random noise, and acquire a reference noise with different phase positions in one cycle; through a certain algorithm, the phase position of the noise-containing signal is matched with the reference noise, thereby subtracting the reference noise of corresponding phase position.

The invention relates to a three-window based common-path interference detecting method and a device and belongs to the field of optical interference detection. According to the method and the device, after parallel lights pass through three windows composed of a to-be-detected object and a rectangular window and then pass through a first lens, a one-dimensional periodic grating and a second lens in turn, the parallel lights are emitted toward an image sensor; the parallel lights are collected by the image sensor and a computer connected with the image sensor, thereby obtaining an interference image containing three patterns; the intensity distributions are respectively I-D / 3, I0 and ID / 3; and the intensity distributions are introduced into a formula, thereby calculating phase distribution of the to-be-detected object. The method and the device have the characteristics of excellent stability, low system complexity and simple phase recovering algorithm.

The invention provides a prestack seismic data random noise suppression method and system, and belongs to the field of geophysical exploration technology and deep learning. According to the pre-stack seismic data random noise suppression method, actual seismic data, pre-stack shot gather forward modeling data and a GPU parallel mode are utilized to obtain an intelligent random noise suppression network, the intelligent random noise suppression network is utilized to process the actual pre-stack seismic data, and pre-stack seismic data after random noise suppression are obtained. According to the method, the U-NET network is applied to seismic data denoising, the high requirement of a conventional convolutional network for the training data volume and the neural network scale is effectively avoided, multi-GPU data parallel network training is achieved, the training efficiency is improved, and effective suppression of prestack data random noise is achieved.

The invention provides a seismic data frequency band expansion processing method based on compressed sensing. The method comprises the steps of: S1), representing a seismic signal based on a reflection coefficient of an underground stratum; S2) converting the seismic signal to obtain a frequency spectrum corresponding to the seismic signal; S3) carrying out discrete operation on the spectrum equation of the frequency spectrum to establish an equation of a reflection coefficient effective selection spectrum; and S4) carrying out optimization solution on the reflection coefficient so as to obtain a broadband sparse pulse reflection coefficient sequence.

The invention relates to a seismic data random noise suppression method and system, and the method is characterized in that the method comprises the following steps: 1) obtaining original seismic data, and carrying out non-negative processing on the seismic data; 2) constructing a denoising model based on self-paced learning and non-negative matrix factorization according to the processed seismicdata; 3) solving the denoising model based on the self-paced learning and the non-negative matrix factorization by adopting an improved block coordinate descent method; and 4) performing inverse operation of non-negative processing in the step 1) on the obtained optimal solution to obtain denoised seismic data. The provided method can be widely applied to the technical field of seismic exploration.