555 results about "Noise estimation" patented technology

The invention discloses a cancer image denoising method and system based on a multi-scale neural network, and the method comprises the steps: carrying out the adaptive segmentation of a cancer image, carrying out the noise estimation and statistical analysis of each segmented region, and automatically adjusting the denoising parameters; performing multi-scale feature extraction by adopting wavelet transform and pyramid decomposition, and analyzing and processing the image after parameter adjustment; performing quality evaluation and distortion detection on the image after multi-scale processing, automatically identifying and positioning artifacts and distortion, and correcting and compensating interference by adopting a point spread function inversion and Wiener filtering method to obtain a corrected image; and after histogram equalization and contrast-limited adaptive histogram equalization are applied to the corrected image, denoising processing is carried out by using a convolutional neural network in combination with sparse representation and a dictionary learning algorithm, and a denoising result is obtained. According to the method, noise in different types of cancer images can be effectively removed, meanwhile, key diagnosis information is reserved, and the image quality and definition are improved.

The invention discloses a weak light image enhancement method fusing noise adaptive diffusion and illumination perception, and the method comprises the steps: firstly designing a reflection-illumination decomposition module RID based on a Retinex theory, and extracting a reflectivity component and an illumination component of an input weak light image; then designing an ADRD (adaptive diffusion reflection denoising) module, performing denoising enhancement on the reflectivity component, and outputting an enhanced reflectivity image; introducing a noise estimation network, and dynamically adjusting the de-noising amplitude of the diffusion model according to the noise intensity of the region in the reflectivity graph; designing an illumination perception enhancement module IEM, fusing an attention mechanism and a local enhancement block, enhancing the illumination image, and outputting an enhanced illumination component; and building a complete neural network framework, carrying out joint training on the provided module, finally testing the stored neural network model, and outputting an enhanced image. According to the method, the problems of detail loss, noise residue and uneven illumination of the enhanced image in the prior art are solved.

The invention relates to an accelerated high-quality MRI (Magnetic Resonance Imaging) image reconstruction method based on a TC-KANReccon model. The method comprises the following steps: S1, collecting and preprocessing image data; s2, an accelerated high-quality MRI image reconstruction model based on a TC-KANReccon model is built; and S3, designing a noise cutting strategy of linear reduction of the cutting threshold. In each step of the sampling process, a linear decreasing cutting strategy is adopted according to the current progress of the sampling step. The strategy is applied to noise estimation generated by an MF-UKAN diffusion trunk so as to generate an MRI image with a better visual effect; s4, repeatedly training a diffusion model and a VAE module by taking the preprocessed MRI image and the full-sampling k-space data as input, optimizing network parameters, and continuously performing iterative optimization; and S5, inputting the under-sampled k-space data into the trained accelerated high-quality MRI image reconstruction model based on the TC-KANReccon model, and applying a matched noise cutting strategy of linear reduction of a cutting threshold to finally generate a high-quality MRI image.

The invention discloses a bimodal neural network and spectral subtraction combined underwater acoustic signal noise suppression method. Comprising the following steps: firstly, constructing a simulation underwater sound data set containing self-noise; thirdly, constructing a bimodal neural network, then training the bimodal neural network until training is completed, and obtaining a platform self-noise estimation model; inputting a to-be-processed noise-containing underwater acoustic signal into the platform self-noise estimation model, and outputting estimation platform self-noise by the model; and finally, based on the self noise of the estimation platform, noise reduction processing is carried out on the noise-containing underwater acoustic signal to be processed by using the improved spectral subtraction method, and then a noise-reduced underwater acoustic signal is obtained. The method combines an advanced deep learning method and a classical spectral subtraction algorithm, has the capabilities of autonomous learning and efficient noise reduction, remarkably improves the signal-to-noise ratio of the underwater acoustic signal, is suitable for noise reduction of the underwater acoustic signal in a non-stationary and complex marine environment, remarkably improves the low-frequency noise suppression capability and the target signal fidelity, and has a good application prospect. The method is suitable for underwater acoustic target detection, underwater acoustic communication and other scenes.

The invention discloses an anti-interference detection method of a wafer temperature measurement system, and belongs to the technical field of semiconductor rapid heat treatment process, the anti-interference detection method comprises the following steps: a temperature sensor detects the temperature of a wafer, an original sampling value detected by the temperature sensor is firstly subjected to noise interference filtering processing through a temperature detector, and then the original sampling value is subjected to noise interference filtering processing; sending the processed data as the detection temperature of the wafer to a temperature controller, comparing the detection temperature with the formula set temperature in the temperature controller, and then adjusting the output power to form closed-loop control; the noise interference filtering processing comprises the steps that an original sampling value obtained by the temperature sensor is subjected to data verification, the verified data is subjected to extreme value removal filtering, then real-time estimation is carried out by adopting a noise estimation algorithm, steady-state noise is identified, and the steady-state noise is removed to obtain the actual temperature of the wafer. According to the method, extreme value removal filtering is adopted in the aspect of random disturbance, a noise estimation mode is adopted in the aspect of stable sine disturbance, and the temperature measurement accuracy is remarkably improved.

According to the VLA-based body robot SLAM method and device and the storage medium, a VLA large model is introduced on the basis of multi-modal fusion SLAM of traditional point cloud geometry, vision and the like, perception is improved from a geometric layer to semantic concept alignment, and the SLAM is more accurate. A VLA large model is used for carrying out dynamic prediction updating on dynamic interference filtering, key frame screening, factor graph relation construction, noise estimation, loopback detection and the like, the real-time requirement is met in the modes of incremental optimization and the like, and the dislocation problem of geometric constraints is corrected through global semantic constraints. The stability of the robot SLAM in extreme scenes such as excessive environmental dynamic interference, loud sensor noise and environmental degradation is improved, and the constructed hierarchical situation map can meet the requirement of a high-order navigation task while the geometric accuracy is met, so that the robot SLAM can be deployed to carriers such as a body-equipped intelligent carrier for subsequent application.

The invention discloses a combined navigation method and system of an unmanned surface vehicle. The method comprises the following steps: forming a GNSS (Global Navigation Satellite System) / INS (Inertial Navigation System) integrated navigation system by using a GNSS (Global Navigation Satellite System) and a strapdown inertial navigation system (INS), and establishing a mathematical model of a GNSS / INS integrated navigation system filter; the mathematical model comprises a state equation and a measurement equation; establishing a mathematical model of the influence of the GNSS auxiliary information on the GNSS positioning precision so as to preliminarily adjust the measurement noise in the measurement equation; through an improved Sage-Husa adaptive filtering algorithm, adaptive estimation is carried out on the preliminarily adjusted measurement noise, and a measurement noise estimation final value is obtained; and according to the motion characteristics of the unmanned surface vehicle, adding motion constraint conditions to the measurement equation. The GNSS auxiliary information mathematical model based on multivariate function fitting is established, the influence of the external environment on the positioning precision is objectively reflected, the measurement noise in the filter is dynamically adjusted, and the positioning precision of the integrated navigation system under the interference condition is remarkably improved.

The invention belongs to the field of medical image processing and radiotherapy quality control, particularly relates to a gamma distribution diagram calculation method based on a diffusion model, and aims to solve the problem that the traditional radiotherapy quality guarantee efficiency is low. The method includes deriving a field flux profile image from a radiotherapy planning system. A reverse process noise estimation network is improved based on a conditional diffusion model, a field flux distribution diagram image is taken as conditional input, field flux distribution diagram information is extracted layer by layer at an encoder part, information fusion with noise path coding is carried out at different coding scales, diffusion noise is predicted by adopting multi-scale fusion Unet, and the prediction precision is improved. And constructing to obtain a gamma distribution diagram generation model. A dual-path residual block is adopted as a basic module of an encoder and a decoder of the multi-scale fusion Unet, a moving window jump connection module is adopted to integrate global and local features of different stages, and the moving window jump connection module introduces spatial offset through a moving window mechanism.

The invention discloses a multivariate time series diffusion interpolation method based on historical data, terminal equipment and a storage medium, and aims to select multiple sections of historical data most matched with a current time period to supplement training data so as to solve the problem of insufficient training data in a self-supervised training process. On the basis, a historical data fusion scheme is further designed, the similarity and the time correlation between the target observation data and the historical data are fully considered, multiple historical data segments are integrated with the target observation data in a normalization mode, and it is ensured that the fused training data are distributed in an approximate value range of a true value. The diffusion process is used for training a noise estimation model, the fused training data and part of observation data are input into the diffusion process at the same time, rich feature information is provided for model training, the denoising process starts from randomly generated noise data, noise is gradually removed through the noise estimation model, and finally interpolation data are obtained.

The invention discloses an image generation method. The method comprises the steps of obtaining a reference face image, a text cue word and source noisy data ZT; the text cue word is used for describing features of the customized image, and ZT is noisy data obtained after Gaussian noise of T time steps is gradually added to the reference face image through a diffusion model; generating a lexical element embedding vector based on the reference face image and the text cue word, and inputting the lexical element embedding vector into a de-noising device of the diffusion model to obtain a de-noising device taking the lexical element embedding vector as a condition; inputting the Zi into a de-noising device under the condition of lexical element embedding vectors for noise estimation to obtain first noise data belonging to i; according to the set i, the Zi and the reference face image, second noise data # imgabs0 # are generated, noisy data Zi-1 of the (i-1) th time step are generated according to # imgabs1 # and Zi, the initial value of i is T, the step is executed circularly after i is subtracted by 1 accumulatively until i is equal to 0, and noise-free data Z0 are obtained; and generating a customized image corresponding to the reference face image based on the Z0.

The invention discloses an image super-resolution method based on a frequency domain residual diffusion model. The image super-resolution method comprises the following steps: selecting a frequency domain wavelet signal of a low-resolution image as the input of a first segment of residual CNN network to generate a high-frequency component and a low-frequency component of an enhanced image; the difference value between the high-frequency component of the enhanced image and the high-frequency component of the high-resolution target image is used for training a noise estimation network in a diffusion model; generating an estimated residual high-frequency component by using a noise estimation network; adding the estimated residual high-frequency component and the high-frequency component of the enhanced image, and then superposing the low-frequency component of the enhanced image to obtain a high-frequency component and a low-frequency component of a final image; and performing inverse discrete wavelet transform on the obtained high-frequency component and low-frequency component to obtain a final super-resolution image. The image super-resolution method is used for converting a low-resolution image into a high-resolution image. A diffusion model is adopted to replace a GAN network, and the training process is more stable.

The invention relates to the technical field of image processing, in particular to an artificial intelligence-assisted CT image quality enhancement and noise reduction method, which comprises the following steps of S1, acquiring low-dose CT scanning image data; s2, preprocessing the acquired low-dose CT image data; s3, extracting preliminary feature information in the CT image; s4, performing adaptive estimation on noise in the image; s5, inputting the noise estimation result and the initial feature information into a deep convolutional denoising network; s6, performing quality enhancement on the denoised image based on a generative adversarial network framework; s7, reconstructing the CT image after quality enhancement; and S8, performing medical applicability detection on the final image. Through the generative adversarial network, the convolutional neural network and the medical image discrimination model, denoising and quality enhancement of the low-dose CT image are realized, the CT image with a high signal-to-noise ratio and high structure fidelity is generated, the medical applicability of the CT image is ensured, and clinical diagnosis requirements are met.

Embodiments described herein are generally directed to an end-to-end trainable degradation restoration network (DRN) that enhances the ability of a super-resolution (SR) subnetwork to deal with noisy low-resolution images. An embodiment of a method includes estimating, by a noise estimator (NE) subnetwork of the DRN, an estimated noise map for a noisy input image; and predicting, by the SR subnetwork of the DRN, a clean upscaled image based on the input image and the noise map by, for each of multiple conditional residual dense blocks (CRDBs) stacked within one or more cascade blocks representing the SR subnetwork, adjusting, by a noise control layer of the CRDB that follows a stacked set of a multiple residual dense blocks of the CRDB, feature values of an intermediate feature map associated with the input image by applying (i) a scaling factor and (ii) an offset factor derived from the noise map.

The invention relates to the technical field of inertial navigation and sensor fusion, in particular to an IMU (inertial measurement unit) attitude estimation and error correction method based on a state space physical information neural network, which comprises the following steps: acquiring original data of an IMU and preprocessing; constructing a state space model, and designing a state transition physical information neural network based on the state space model; training the state transition physical information neural network based on the preprocessed original data, optimizing network parameters through a joint loss function in the training process, and obtaining the trained state transition physical information neural network; and collecting real-time data streams of the IMU, inputting the real-time data streams into the trained state transition physical information neural network, and outputting the corrected pose, position and speed. According to the method, physical constraints and adaptive noise estimation are fused, so that the attitude estimation precision and the error correction capability in a dynamic complex environment are remarkably improved.

The invention discloses an adaptive edge detection iterative denoising point cloud reconstruction method and system. A regularization parameter and a kernel bandwidth are automatically set through point cloud normalization and local covariance noise estimation; an implicit curved surface is fitted based on a parameterized Gaussian field, and initial curved surface construction without a normal direction is realized. Edge detection is carried out by utilizing Voronoi covariance measurement, and the edge detection is continuously mapped into a projection weight, so that feature details of the point cloud are reserved in an acute angle region, and smooth denoising is realized in a flat region. The regularization intensity is dynamically adjusted through residual trend analysis to enhance detail representation, and a triangular mesh is extracted and output in combination with octree discretization and contour surface extraction. The method has unsupervised adaptivity, has outstanding performance in the aspects of edge fidelity, noise robustness and complex structure adaptability, can effectively process point cloud data with high noise, missing normal direction or irregular structure, and is suitable for the fields of engineering survey, reverse modeling, cultural relic digitization and the like.

The invention discloses a video noise estimation method, a video noise estimation device and a computer storage medium. The video noise estimation method comprises the following steps: calculating an inter-frame difference result between corresponding pixel points in adjacent frame images; performing image edge calculation on each pixel point in each frame image to obtain an edge detection feature value corresponding to each pixel point in each frame image; determining a static edge point in each frame image according to the edge detection feature value corresponding to each pixel point in each frame image; and determining a video noise estimation value according to an inter-frame difference result between the static edge point in each frame image and each corresponding pixel point in the adjacent frame image. By adopting the method, the influence of image texture details and inter-frame motion on the video noise estimation result can be avoided, so that the video noise estimation result can better reflect the real noise level.

The invention provides a signal recovery method integrating noise detection and prior estimation, which comprises the following steps of: acquiring pre-estimation of an original signal by utilizing a zero interference equalization module based on an observation signal and a channel matrix; carrying out iterative optimization on the estimated value of the signal by utilizing a self-adaptive signal processing module; based on pre-estimation of the original signal and information extracted from noise observation of the original signal generated by the adaptive signal processing module, pre-estimation of signal distribution is obtained through a semi-parameter Bayesian module, and estimation of the original signal distribution is updated in each iteration process of the adaptive signal processing module; and the estimated value of the noise variance is updated through the iterative noise estimation module, and is updated in each iteration process of the adaptive signal processing module. According to the method, the difficulty of inaccurate signal recovery caused by noise fluctuation brought by environmental factors is overcome, and meanwhile, the signal can be accurately recovered when the prior information of the original signal is lacked.

A label noise estimation method based on a manifold regularization transfer matrix provided by the present invention pre-trains a first network in a second network, and after distilling a data set, inputs the obtained sub-data set into the second network to obtain the probability of the class to which the data instances in the sub-data set belong and obtain a transfer matrix related to the data instances; further calculates the cross-entropy loss of the second network according to the data instance labels, and combines an association matrix expressing the consistency of the data instances belonging to the same manifold and a penalty matrix of the data instances belonging to different manifolds to calculate the loss function of the second network; adjusts the loss function to reduce the training of the second network to obtain a trained second network, thereby completing the estimation of the class to which the data instances belong. The present invention can reduce the estimation error without affecting the approximation error of the transfer matrix, and experiments prove that the present invention can achieve excellent performance in label noise learning.

The invention provides an adaptive robust dynamic state estimation method for a power system based on hybrid measurement, which comprises the following steps of: firstly, realizing fusion of hybrid measurement through a measurement transformation technology, and constructing pseudo measurement based on space-time characteristics of the system to form a hybrid measurement set; furthermore, an adaptive forgetting factor is introduced, and a noise estimation algorithm is improved, so that the time-varying system noise can be estimated more quickly and accurately, and the dynamic tracking capability of the algorithm is improved; secondly, estimating measurement noise by analyzing characteristics of the measurement device, and constructing a robust factor based on standard innovation to suppress bad data; and finally, dynamic state estimation is carried out based on an extended Kalman filtering criterion, and the accurate operation state of the system can be obtained under the influence of unknown model parameters, system state change and bad data.

The invention relates to the technical field of audio processing, in particular to an audio optimization method and device based on artificial intelligence, computer equipment and a medium. According to the method, a linear layer is used for mapping a spectrum envelope of a to-be-optimized audio to obtain an envelope feature, an embedding layer is used for embedding a standard audio parameter into a parameter feature, a prediction model is used for predicting a fusion feature of the envelope feature and the parameter feature to obtain a prediction pitch curve, and a noise adding model is used for adding noise to a Mel spectrum of the to-be-optimized audio to obtain a predicted pitch curve. Calculating the noise of the noise addition result by using a noise estimation model to obtain predicted noise, updating the noise estimation model according to the predicted noise, the real noise and the predicted pitch curve, calculating the reference noise of the noise addition result by using the updated noise estimation model, and de-noising the noise addition result according to the reference noise. The noise estimation model is optimized in combination with pitch information, so that the denoising process meets the pitch requirement, and the audio optimization effect is improved.

The invention relates to the technical field of cable force detection, in particular to a bridge inhaul cable fundamental frequency identification method, device and equipment, a storage medium and a product, vibration acceleration data of a bridge inhaul cable structure to be monitored in the duration time is converted into a power spectrum data set, a noise floor is calculated through the power spectrum data set, the noise estimation precision is improved, and the noise estimation accuracy is improved. Energy peak value screening is carried out on the power spectrum data, part of energy peak value data is screened out, a frequency-energy peak value array is obtained, energy evaluation and scoring are carried out on the frequency-energy peak value array and a predetermined candidate fundamental frequency through a preset fundamental frequency scoring model, and the energy evaluation and scoring precision of the candidate frequency is improved; finally, the actual fundamental frequency of the to-be-detected inhaul cable is determined through the score value of the candidate fundamental frequency, manual screening is not needed, the identification efficiency is improved, and meanwhile high-precision identification of the fundamental frequency of the inhaul cable is achieved.

The invention provides an inertia / satellite / vision adaptive integrated navigation method and system. The method comprises the following steps: carrying out inertia measurement and navigation calculation to obtain inertia data; calculating satellite navigation observed quantity according to the satellite navigation data and the inertial data, and constructing a satellite observation model; calculating visual navigation observed quantity according to the visual navigation data and the inertial data, and constructing a visual observation model; aiming at a satellite and a visual observation model, respectively adopting an innovation-based adaptive covariance estimation method to obtain satellite and visual observation noise covariance updated values; constructing a satellite and visual navigation health degree, and calculating a satellite and visual fusion weight according to the satellite and visual navigation health degree; controlling updating of satellite and visual navigation observed quantity according to the satellite and visual fusion weight; and calculating a weighted equivalent observation matrix and a noise covariance, and carrying out filtering estimation. According to the method, an adaptive noise estimation and sensor health degree evaluation mechanism is introduced, dynamic weighted fusion of multi-sensor data is realized, and the robustness and precision of a navigation system in complex environments such as satellite signal lock losing and visual feature missing are improved.

The invention relates to the technical field of navigation positioning, and discloses a satellite positioning method and device based on deep learning and filter coupling driving, and a medium. Determining a coarse positioning result of the receiver based on a least square method by using an observation pseudo-range between the receiver and each satellite and a model pseudo-range between the receiver and each satellite obtained by modeling; extracting observation quality features based on a coarse positioning result; inputting the observation quality features into a noise estimation model obtained based on deep learning to obtain a measurement noise covariance matrix and a noise compensation value; determining a state vector of a filter at an initial moment based on the coarse positioning result, and updating the state vector of the filter step by step based on the measurement noise covariance matrix and the noise compensation value to obtain a state vector at a to-be-estimated moment, the state vector at the to-be-estimated moment comprising a positioning result of the receiver at the to-be-estimated moment; noise and errors in the environment can be effectively compensated, so that the influence of the noise and the errors is effectively reduced, and the accuracy of satellite positioning is improved.

The invention discloses a semantic constraint adversarial sample generation method and system. According to the method, firstly, a confrontation sample generation task based on a natural language instruction is constructed, and parameters are initialized; then deploying a proxy model and a diffusion model for double-branch noise estimation; by judging instruction complexity, mask guidance is selectively adopted to realize natural constraints of spatial differentiation; by constructing a residual-guided antagonistic DDIM sampler and combining an adaptive optimization iterative algorithm, the calculation complexity is reduced, and the attack mobility is enhanced at the same time; and aiming at a three-dimensional generation requirement, a three-dimensional Gaussian sputtering rendering model is further integrated, and geometric consistency is ensured through multi-view gradient averaging. According to the method, the problems of inaccurate semantic control, weak migration aggressiveness, poor three-dimensional generation consistency and the like in the prior art are effectively solved, the attack success rate and the visual naturalness are remarkably improved on multiple target models, and an efficient and reliable red team test tool is provided for security evaluation and alignment of a multi-modal large model.

The invention discloses an underwater acoustic OFDM impulse noise estimation enhancement method based on distributed compressed sensing, and relates to the technical field of underwater acoustic communication. Carrying out frequency offset correction on the received OFDM signal by adopting a plurality of carrier frequency offset compensation values to obtain a plurality of groups of related signals; constructing an orthogonal projection matrix, projecting a received signal to a specific orthogonal subspace to eliminate channel components, and separating impulse noise; constructing a joint observation matrix by using the time domain sparsity of the impulse noise and the strong correlation under different carrier frequency offset compensation; and carrying out joint sparse reconstruction on the multiple groups of observation signals by adopting a distributed compressed sensing algorithm, and estimating an impulse noise time domain signal. A distributed compressed sensing algorithm is adopted, two characteristics of impulse noise are utilized, impulse noise estimation accuracy is improved, an impulse noise estimation effect is enhanced, and the problem that impulse noise estimation accuracy is limited under the condition of limited pilot frequency quantity is solved.

The invention belongs to the technical field of voice noise reduction. The invention provides an edge end real-time voice noise reduction method. According to the embodiment of the invention, pre-emphasis, framing and windowing processing is carried out on the input voice signal, so that the high-frequency component of the signal is improved and the distortion of the signal is reduced. And performing fast Fourier transform on the preprocessed input voice signal to convert a time domain signal into an original frequency domain signal. And calculating the Mel frequency cepstrum coefficient of the original frequency domain signal to extract the MFCC characteristic parameter of the speech. Noise estimation is performed on the MFCC characteristic parameters by using the lightweight separable convolutional neural network, so that the parameter quantity and the calculation quantity are reduced, and efficient noise estimation is realized. And performing adaptive iterative Wiener filtering by using the generated noise estimation, and gradually eliminating noise in the voice signal to obtain a filtered frequency domain signal. And inverse fast Fourier transform is carried out on the filtered frequency domain signal, and a time domain signal is recovered. And outputting a time domain signal which is a voice signal after noise reduction.

The invention provides a non-uniform haze degraded image restoration method and a related device, and belongs to the technical field of image processing. The method comprises the following steps: estimating noise randomly sampled from standard normal distribution by using a trained noise estimator to obtain estimated noise, and removing the estimated noise from the noise randomly sampled from the standard normal distribution to obtain a noisy image; constructing a visual state space mixed scale expansion convolution module, and constructing a conditional diffusion U-shaped denoising model based on the visual state space mixed scale expansion convolution module; and based on the guide image, estimating and denoising the noisy image by using a conditional diffusion U-shaped denoising model in an iterative calculation mode to obtain a restored non-uniform haze degraded image. According to the method, the problem that the restoration sharpness degree of the degraded image is poor when the haze concentration change range is large or the non-uniform change is severe is solved.

The invention provides a filtering method and device for channel estimation, electronic equipment and a storage medium. The method comprises the following steps: acquiring initial channel estimation of a reference signal, and determining optimal noise power in a current target channel scene based on the initial channel estimation; performing first-stage Wiener filtering on the initial channel estimation to obtain first filtering channel estimation; and performing second-stage Wiener filtering on the first filtering channel estimation based on the optimal noise power to obtain second filtering channel estimation. Therefore, according to the scheme, the optimal noise power in the target channel scene is acquired in advance, so that the second-stage Wiener filtering can be immediately carried out after the first-stage Wiener filtering is completed, the process of carrying out noise estimation on the reference signal after the first-stage filtering can be omitted, the processing waiting time is reduced, and the processing efficiency is improved. And the processing time delay is reduced, so that the two-stage filter circuit can be designed in a tight coupling manner to enhance the stability of the filter performance.

A method for reducing voice noise, including: acquiring an algorithm activity detection result corresponding to a current audio frame to be processed by detecting the current audio frame using a predetermined voice activity detection algorithm; acquiring a target activity detection result corresponding to the current audio frame by merging a model activity detection result corresponding to a previous audio frame and the algorithm activity detection result corresponding to the current audio frame, wherein the model activity detection result is outputted by a predetermined voice noise reduction network model; acquiring an initial noise reduction audio frame by performing, based on the target activity detection result, noise estimation and noise elimination on the current audio frame; and outputting a target noise reduction audio frame and a model activity detection result corresponding to the current audio frame by inputting the initial noise reduction audio frame into the predetermined voice noise reduction network model.

The invention provides a robot action generation method and a robot control system, and the method comprises the steps: obtaining multi-modal observation data at a current moment; the modal weight and the initial noise trajectory of each modal are acquired; and according to the multi-modal observation data, the modal weight of each modal and the noise estimation model corresponding to each modal, performing iterative exploration on the initial noise trajectory according to a time inverse sequence, and generating an action trajectory sequence of the robot. According to the method, noise estimation of the pre-trained single-mode diffusion strategy can be combined during reasoning, and information of different modes can be efficiently fused, so that on the premise of avoiding complexity of multi-mode data collection and joint training, multi-mode strategy fusion is realized, and meanwhile, the multi-mode data collection and joint training efficiency is improved. And the flexibility, the real-time adaptability, the robustness and the expandability of the action generation process are enhanced. Moreover, by avoiding multi-modal joint training, the training cost and the computing resource demand are greatly reduced.