24 results about "Spectral amplitude" patented technology

The invention provides an infrared image super-resolution method based on a conditional diffusion model, and belongs to the technical field of infrared image super-resolution reconstruction. The method comprises the following steps: adding noise to a high-resolution infrared image to construct a diffusion process; splicing the noisy image, the low-resolution infrared image and the visible light image in a channel dimension, and inputting the spliced image into a U-Net encoder to extract multi-scale fusion features; the decoder recovers the spatial resolution through layer-by-layer up-sampling in combination with jump connection, and meanwhile, structural features of a low-resolution image and texture features of a visible light image are introduced into each layer to carry out spatial condition modulation, so that cross-modal stable guidance is realized; and constructing a joint objective function including denoising loss, frequency domain phase consistency loss and spectrum amplitude constraint loss, and optimizing network parameters through back propagation. According to the method, complementary information of infrared and visible light images can be fully fused, and the problems of insufficient detail recovery and artifact generation of an existing method are effectively solved.

The invention discloses a fabric texture fidelity rendering method and system based on a scanned image and a target color number, and solves the technical problems of poor color perception consistency and texture detail loss. The method comprises the steps of preprocessing a standardized input image through illumination compensation, performing double-space color mapping in an LAB color space, partitioning the image by adopting K-means clustering according to texture features of a gray-level co-occurrence matrix, establishing a color mapping function containing global mapping transformation and local adaptive adjustment, and adaptively determining a fusion coefficient according to regional variance; decomposing the image into a base layer and a detail layer by adopting guide filtering, and performing spectrum constraint on the detail layer to ensure that the spectrum amplitude is consistent with the original; extracting material parameters such as a normal map, a roughness map and a metallity map; according to the method, fabric texture details are completely reserved while color perception consistency is guaranteed, rendering reality is improved, batch processing of multiple groups of target color numbers is supported, and the method is suitable for application scenes such as costume design, e-commerce display and virtual fitting.

The invention provides a real-time speech enhancement method and system based on dual-stage spectral subtraction and dual-mask fusion, and relates to the technical field of speech signal processing, and the method comprises the steps: respectively carrying out the framing, windowing and short-time Fourier transform of a left channel mixed signal and a right channel noise reference signal, obtaining a complex frequency spectrum and an amplitude spectrum of the left channel signal and an amplitude spectrum of the right channel noise reference signal; and performing noise estimation by adopting a first noise multiplication factor based on the amplitude spectrum of the right channel noise reference signal to obtain a noise estimation spectrum, and performing constraint spectrum subtraction on the amplitude spectrum of the left channel signal to obtain voice amplitude estimation of a first stage. According to the method, effective suppression of TTS noise and real-time speech enhancement are realized through framing windowing and frequency domain conversion in combination with over-estimation spectrum subtraction and double-mask fusion through two-stage gain application and time domain reconstruction.

The application discloses a kind of vehicle diesel engine piston ring wear state monitoring method, storage medium and device, it is related to diesel engine state monitoring field, the method includes defining the measured point sensitivity characteristic value for representing the attenuation amount of measured point energy relative to acoustic emission source energy, and determining multiple acoustic emission source positions and measured point positions for installing acoustic emission sensor on cylinder cover;Drive acoustic emission sensor at acoustic emission source to produce excitation, the measured point sensitivity characteristic value between each measured point and acoustic emission source is calculated, determine the measured point for carrying out fault identification among all measured points, as detection measured point;Acoustic emission signal of detection measured point is obtained when piston is located in set position during diesel engine operation, the PSD spectral amplitude of acoustic emission signal is calculated, and compared with the fault threshold under different fault states of piston ring, realize piston ring fault determination.The application can realize the accuracy and effectiveness of piston ring fault diagnosis.

The invention discloses a screen shooting image moire quantification method and device and a storage medium, which are used for improving the compensation precision of a display screen. Acquiring a sub-pixel unit image of the display screen to be tested; performing brightness compensation processing on the sub-pixel unit image to generate an expected brightness compensation image; generating a compensation graph frequency spectrum amplitude image according to the brightness compensation image; generating a stripe frequency band sensing image according to the visual distance and the pixel pitch corresponding to the target color channel; carrying out region division on the frequency spectrum amplitude image of the compensation graph; standard deviations of the elliptical region and the plurality of elliptical ring regions are calculated, and a region standard deviation set is generated; filling a blank image with the standard deviation in the regional standard deviation set to generate a first elliptical ring standard deviation image; performing frequency domain filtering on the first elliptical ring standard deviation image by using the stripe frequency band sensing image to generate a first frequency domain filtering image; and generating a first moire quantization value corresponding to the brightness compensation image according to the compensation image frequency spectrum amplitude image and the first frequency domain filtering image.

The invention relates to the technical field of intelligent sensors, in particular to an insulator defect evaluation method based on vibration acoustics, and the method comprises the steps: analyzing the amplitude difference between each non-backbone component and an acoustic signal, and the discrete degree of all amplitudes in each non-backbone component, determining the noise influence degree according to the number difference between all the non-backbone components and all the backbone components; determining a signal characteristic value based on the proportion of the average frequency spectrum amplitude of each backbone component in the average frequency spectrum amplitude of the acoustic signal and the random fluctuation degree of all the amplitudes in each backbone component, and determining the disturbance degree in combination with the sound influence degree; and based on the confusion degree and the disturbance degree of the acoustic signal, denoising the acoustic signal by adopting a wavelet function so as to evaluate the defect of the insulator. According to the invention, through adaptive denoising, environmental noise interference is suppressed, and the accuracy and reliability of insulator defect evaluation are improved.

The invention discloses a frequency estimation method for judging an interpolation direction based on a phase difference, which belongs to the technical field of radar signal processing and comprises the following steps of: performing FFT (Fast Fourier Transform) on an original signal of an LFMCW (Linear Frequency Modulated Continuous Wave) radar transmitting signal, finding out positions of a maximum value and a second maximum value corresponding to a frequency spectrum amplitude, solving a phase at a spectral line, comparing the phase difference and judging the interpolation direction. According to the method, the Rife algorithm is improved, interpolation direction judgment is carried out through the phase difference between the maximum spectral line and the second maximum spectral lines adjacent to the two sides, better anti-noise performance is achieved, relatively high stability is achieved, and the problem of interpolation direction misjudgment is avoided.

The invention discloses a gas leakage detection method and system based on laser speckle imaging, and the method comprises the steps: irradiating the outer wall surface of a to-be-monitored region with laser beam expansion to form laser speckles, and continuously collecting a plurality of monitoring speckle images of the to-be-monitored region; segmenting the monitored speckle image and the corresponding original speckle image without leakage into a plurality of sub-windows, and calculating the frequency spectrum amplitude and the frequency spectrum phase of each sub-window; calculating a correlation coefficient between the two sub-windows according to the spectrum amplitudes and the spectrum phases of the sub-windows corresponding to the original speckle image and the monitoring speckle image; marking the sub-windows in the monitoring speckle image corresponding to all the correlation coefficients smaller than a preset threshold value as gas leakage coarse positioning areas; and for each pixel point in the leakage coarse positioning area, calculating a pixel-level imaging operator of the pixel point, and taking the pixel point corresponding to the maximum pixel-level imaging operator as a gas leakage point.

The application discloses a kind of traceable pressure sensor dynamic response calibration method based on absorption spectrum reference pressure input, in shock tube is driven section tail end wall position arrangement to be calibrated pressure sensor, and set optical measurement window in reflected shock action area, synchronous acquisition pressure sensor output signal and absorption spectrum signal;Absorption spectrum signal is preprocessed and spectral line fitting, calculate the gas temperature and collision broadening coefficient after reflected shock wave, and reference pressure amplitude P5 after reflected shock wave is obtained accordingly;Pressure amplitude P1 before incident shock wave is measured using pressure gauge after metrological calibration;According to P1, P5 and pressure mutation time t0 Reference pressure step input signal based on Heaviside step function is constructed;Reference pressure step input signal and pressure sensor output signal are processed, and the ratio of the spectral amplitude of the two is used as the sensitivity of the pressure sensor at the corresponding frequency;Error source in the process of calculating reference pressure amplitude, pressure mutation time and sensitivity is evaluated and calculated, and the sensitivity calibration result of pressure sensor is obtained.The application can obtain the sensitivity characteristics and its uncertainty of pressure sensor.

The invention discloses a hyperspectral diffraction imaging device, system and method based on blind coding, and relates to the technical field of hyperspectral coherent diffraction imaging.The method comprises the steps that a hyperspectral light beam emitted by a light source is transmitted to a mask plate through a reflector after being subjected to beam expanding, focusing and collimation; controlling the mask plate to move from the first position to the second position, so that the mask plates at different positions before and after movement have overlapped areas along the light path direction, and obtaining a diffraction image of each lamination scanning position; and according to the diffraction image, performing joint reconstruction on the unknown mask plate, the spectrum and the imaging object by using a broadband phase reconstruction algorithm based on multi-parameter constraint to obtain amplitude and phase information of the reconstructed mask plate, the spectrum and the imaging object so as to complete hyperspectral illumination coherent diffraction imaging. According to the method disclosed by the invention, multi-spectrum amplitude-phase-spectrum joint reconstruction of the hyperspectrum under low-time coherent illumination can be realized by utilizing laminated scanning under the condition that codes are unknown.

A method for recovering missing spectral amplitude and phase information of an input real, positive valued image or signal possibly carried by incoherent or coherent waves.

The invention provides a spectrum quantitative detection model updating method applied to a recessive decision boundary, which comprises the following steps of: (1) collecting spectrum data of an unmarked sample, and obtaining frequency characteristics of a spectrum; (2) calculating the distance between the unmarked sample and the modeling sample on the spectral amplitude characteristic and the spectral frequency characteristic; (3) selecting a valuable sample through a statistical inference algorithm based on the calculated distance between the two types of features; (3) selecting representative samples from the valuable samples by using hierarchical clustering and local linear reconstruction methods, and manually marking label values of the representative samples; and (4) designing a target function for model updating, and updating the model by adopting a batch training method based on an information extraction ratio. The method focuses on the updating problem of the spectrum quantitative detection model, the representative sample is obtained based on the implicit decision boundary mode, and rapid, reliable and low-cost updating of the model is achieved.

The invention discloses a multi-domain fusion voice noise reduction method and system and a computer readable storage medium, and relates to the technical field of voice enhancement and signal processing, and the method comprises the following steps: recognizing a candidate line spectrum through constructing a power spectrum peak value comprehensive score, and carrying out the adaptive notch; performing compensation in combination with wavelet adaptive threshold processing and improved logarithmic spectrum amplitude estimation; and weighted fusion is further carried out on the signals after wave trapping and compensation through spectrum energy entropy. Rapid suppression of narrowband line spectrum interference is realized through spectrum line peak detection and an adaptive notch technology, broadband background noise is effectively eliminated by means of a wavelet coefficient energy ratio eta adaptive threshold algorithm and an OM-LSA spectrum compensation technology, and recursive robust estimation in a time-frequency domain is completed by adopting frequency domain three-dimensional Kalman filtering. And finally, signal detail reconstruction and spectrum equalization are realized by using a lightweight residual CNN network without training weight, so that the optimal balance of high gain and low distortion is achieved in a complex noise environment.

The application discloses a single-station synthetic aperture passive positioning method and system based on a maximum spectral sharpness criterion, and belongs to the fields of signal processing technology and passive positioning. The method comprises the following steps: controlling a single motion platform to continuously observe a non-cooperative radiation source and collect a time-domain baseband data sequence; constructing a virtual detection grid in a geographical space, and generating a corresponding geometric de-rotation operator for each grid node; performing point-by-point phase stripping processing on the baseband data by using the operator to obtain a residual phase sequence; mapping the residual phase sequence to a frequency domain, calculating an energy concentration norm of a spectral amplitude as a decision statistic of spatial matching; and determining the spatial coordinates of the radiation source and simultaneously inverting a transmitting-receiving frequency deviation by searching for a global maximum value point of the statistic. The application utilizes the physical characteristics that the signal frequency energy is highly concentrated under the spatial matching state, effectively avoids the defocusing problem of a traditional time-domain coherent accumulation algorithm under a frequency non-synchronous condition, and significantly improves the precision and robustness of single-station passive positioning.

This invention discloses an InSAR phase filtering method based on complex interferograms, comprising: preprocessing acquired primary and secondary SAR images to obtain complex InSAR interferograms; dividing the complex InSAR interferograms into square-shaped interferogram patches along the azimuth and range directions; converting each interferogram patch from the spatial domain to the frequency domain signal, constructing a corresponding spectral amplitude map; using K-means clustering to extract an initial low-pass filter from the spectral amplitude map, and constructing a phase filter together with a Kaiser window function; passing the frequency domain signal of the interferogram patch through the phase filter, where the signal frequency power is preserved, noise at high-frequency positions is filtered out, and noise at low frequencies is smoothed out, resulting in a filtered frequency domain signal; converting the filtered frequency domain signal back to the spatial domain to obtain filtered interferogram patches, which are then stitched together to obtain a filtered InSAR interferogram. This invention contributes to the promotion and advancement of current time-series InSAR interferometry technology and its engineering applications.

In accordance with various embodiments, a method for creating a training dataset for an artificial intelligence-based arc-fault circuit interrupter is provided. In some embodiments, the method comprises collecting a number of frames of no-arc data, calculating a spectral amplitude value for each frame of no-arc data and summing the spectral amplitude values; calculating an average of the sums of the spectral amplitude values of the no-arc data; collecting a number of frames of arc data; calculating a spectral amplitude value for each frame of arc data and summing the spectral amplitude values; for each frame of arc data, comparing the sum of the spectral amplitude values to a threshold based on the average of the sums of the spectral amplitude values of the no-arc data; and adding each frame of arc data whose sum of the spectral amplitude values exceeds the threshold to an AI data model training dataset.

The present application belongs to the technical field of neural signal processing, and specifically relates to a neuron spike signal data enhancement method based on VAE-GAN. The present application comprises: constructing an improved variational autoencoder generative adversarial network, integrating a time attention module in the decoder of the generative network to capture and reconstruct the key time dynamic characteristics of the neuron spike signal; introducing a double discriminator architecture for evaluating the fidelity of the generated signal in the frequency domain and suppressing high-frequency artifacts; designing a composite loss function containing a spectral amplitude loss to impose a time-frequency double constraint on the generator during the adversarial training process, ensuring that the generated samples are highly consistent with the real signal in terms of time domain morphology and frequency domain characteristics; and synthesizing high-fidelity spike waveforms in batches for a small number of neuron classes, which are used to expand the original data set and construct a class-balanced training set. Experiments on multiple data sets and downstream classification models prove that the present method can improve the classification performance of neuron spike signals and has high efficiency and universality.

A method includes steps for creating at least two replicas of an input pulse to be characterised, varying the relative amplitude of the two replicas within a range, creating a nonlinear signal at each case of said amplitude variation, measuring the spectra of the nonlinear signals and recovering the spectral amplitude and phase of the input pulse with a proper algorithm. The system includes a replicator for creating at least two replicas of the input pulse and varying their relative amplitude within a range of relative amplitudes, a nonlinear medium, which obtains a nonlinear signal for each relative amplitude, and an analyzer, associated to the nonlinear signal, for measuring and characterising spectrally each nonlinear signal.

The present application belongs to the technical field of digital audio processing, and provides a classification-based sibilance suppression method to solve the problem that the prior art causes perceptible loss of useful speech information when suppressing sibilance due to the lack of fine-grained distinction between sibilance and useful speech. The present application includes two stages of classifier construction and sibilance suppression. The spectrum of a training audio frame is extracted, the spectral amplitude data of the sibilance frequency band is grouped, the sum of the spectral amplitudes in each group is calculated, and the amplitudes of each group are spliced to form a feature, on the basis of which a sibilance classifier is trained. In sibilance suppression, the sibilance type probability output by the sibilance classifier is used to perform fine-grained type prediction on the audio data to be processed in units of frames, and the prediction results of consecutive frames are smoothed through median filtering. Finally, the relative energy of sibilance and non-sibilance is calculated to attenuate the sibilance component, thereby effectively suppressing the sibilance in the audio while maintaining the original speech information.

The application belongs to the technical field of distributed optical fiber sensing, and relates to an anti-distortion distributed optical fiber sensing method and system based on array phase spectrum demodulation. A series of reflection peaks with a signal-to-noise ratio RBS higher than that of a reflection peak are generated by using a super-weak reflection fiber Bragg grating array as a sensing optical fiber, a reflection peak sequence is formed, and the spectral amplitude of a beat frequency signal is increased. The discrete beat frequency signal with frequency domain distortion output by an OFDR sensing system is converted to the frequency domain by Fourier transform with a symmetric window function to obtain the corresponding discrete frequency spectrum, the amplitude, phase and other characteristic parameters of the reflection peak sequence in the frequency spectrum are tracked and extracted, the frequency correction factor of the reflection peak sequence at all measurement periods is calculated by using an amplitude interpolation correction method, the corrected accurate array phase spectrum is obtained, and phase difference and strain calculation are performed, so that multi-point quasi-distributed strain demodulation for suppressing frequency domain distortions such as Doppler frequency shift is realized, the frequency resolution limitation caused by the fence effect is broken, and the fidelity of the frequency domain distortion signal is restored.

The invention discloses a speech synthesis method suitable for an off-line scene of the Internet of Things, and relates to the technical field of speech data processing. The method comprises the following steps: acquiring an original human voice syllable sample and a to-be-synthesized target syllable sequence; calculating short-time root-mean-square energy of each frame of the original voice, and performing endpoint detection, framing windowing and FFT (Fast Fourier Transform) to obtain a frequency domain spectrum; extracting an amplitude higher than a preset multiple of the average amplitude in each frame of spectrum and a corresponding frequency as a characteristic wave peak, and packaging the characteristic wave peak into a compressed characteristic data packet to be stored locally; extracting a compressed data packet corresponding to the target syllable from the local, taking the characteristic wave crest of each frame as a node, complementing the frequency spectrum amplitude between nodes through linear interpolation, and adjusting the amplitude according to the short-time energy of each frame to obtain a time domain voice frame; and carrying out frame splicing and tone adjustment, and outputting high-quality human audio voice. According to the invention, while the low storage requirement of speech synthesis is met, the requirements of instantaneity and clear tone quality are met.

This invention relates to the field of ultrasonic flow measurement technology. This application discloses a relative time-of-flight estimation method based on an ultrasonic flow meter. The method includes: aligning the uplink and downlink echo signals in the time domain using prior absolute time of flight to prevent phase skipping; performing FFT transformation and frequency domain conjugate multiplication on the signals to extract amplitude and phase information; constructing a cost function using spectral amplitude weighting; and estimating the residual delay through least-squares fitting to finally obtain a high-precision relative time of flight. This method performs correlation operations in the frequency domain, significantly reducing computational complexity, improving estimation accuracy and noise immunity, and is suitable for embedded low-power systems.

The application discloses an adaptive threshold SAMP reconstruction method of power quality disturbance signals. The method obtains compressed observation values by constructing a random Gaussian observation matrix, and uses a discrete Fourier transform to perform sparse representation on an original signal. In the iterative reconstruction process, the application first introduces the spectral amplitude difference as an adaptive termination basis, and combines a dynamic threshold updating function to realize automatic adaptation to different noise levels and different disturbance characteristics. The method effectively avoids the problems of high estimation of traditional SAMP sparsity and redundant iteration, and significantly reduces the calculation load. Compared with the OMP, original SAMP and other methods, the application method can reduce the iteration number by 30% to 60%, the reconstruction signal-to-noise ratio is increased by 2 to 5 dB, the root mean square error is reduced by 10% to 25%, and the application method has higher robustness and real-time performance in the reconstruction of power quality disturbance signals, and is very suitable for compressed sampling, edge computing and high-speed signal reconstruction scenes in a power quality monitoring system.

The invention discloses a model training and interference suppression method. The model training method comprises the following steps: acquiring a complex baseband time domain sample; decomposing the complex baseband time domain sample into a real part channel and an imaginary part channel to obtain a real part channel signal and an imaginary part channel signal; analyzing the dual-channel input tensor by using a neural network model to obtain a dual-channel output tensor, and performing reverse normalization processing on the dual-channel output tensor to obtain a predicted interference component; determining time domain loss, frequency domain spectrum amplitude loss and residual energy penalty loss according to the real interference component and the predicted interference component corresponding to the complex baseband time domain sample; determining a total loss function; and training the neural network model based on the total loss function until a preset stop condition is met. According to the method and the device, the technical problem that interference cannot be effectively suppressed due to the fact that the interference component in the complex baseband time domain signal cannot be accurately predicted in the prior art is solved.