36 results about "Adaptive smoothing" patented technology

An apparatus comprising an interface and a processor. The interface may be configured to receive pixel data of an environment. The processor may be configured to process the pixel data arranged as video frames, generate a luminance distribution map of the video frames in response to a low-pass filter operation, determine a plurality of defogging intensity weights for the luminance distribution map, perform adaptive smoothing to each of the plurality of defogging intensity weights, and generate defogged video frames in response to the video frames and the plurality of defogging intensity weights with the adaptive smoothing. The plurality of defogging intensity weights may each correspond to one of a plurality of luminance intervals of the luminance distribution map. The adaptive smoothing may be configured to prevent brightness differences in the defogged video frames.

The application discloses a multi-agent collaborative perception feature enhancement method based on context aggregation, relates to the technical field of automatic driving and intelligent networked vehicles, and comprises the following steps: acquiring current frame features and at least one frame of historical features; adaptively aligning the historical features based on a motion prediction network and a deformable convolution offset; generating semantic consistency weights and identifying time sequence discontinuous regions based on a correlation prediction network; fusing the current and historical features according to the weights and carrying out state space selective scanning, pooling, denoising and aggregation; generating multi-scale features, combining scene complexity, quality evaluation and time sequence consistency constraints to determine scale weights and then fusing; and inputting an LSTM to perform time sequence modeling and output enhanced context features. Invalid alignment is avoided through motion compensation and semantic consistency constraints, noise and scale jitter are suppressed through position-level denoising and multi-scale adaptive smoothing, and the collaborative perception robustness and detection accuracy are improved.

The application belongs to the technical field of image processing, and relates to an extruded pipe surface defect detection method and system based on machine vision. The method acquires an original gray image of the extruded pipe surface, extracts texture energy response by using a multi-direction filter set, acquires local maximum, minimum and global average texture energy response values; constructs a ratio operation according to the response values to calculate a local spatial anisotropy index; utilizes the nonlinear attenuation characteristics of an exponential function to dynamically map a basic scale constant based on the local spatial anisotropy index, and calculates a dynamic smoothing scale of each pixel point; uses a Gaussian kernel function corresponding to the dynamic smoothing scale to perform adaptive smoothing processing on the image and constructs a Hessian matrix, solves eigenvalues, and combines a reference background curvature constant to calculate a confidence enhancement response, so as to extract real crack defects. The application realizes noise suppression and weak crack reservation, and improves the accuracy of pipe surface defect detection.

A large language model quantization method based on channel arrangement and activation adaptive smoothing belongs to the technical field of large language model compression and efficient inference. In order to solve the problem that the existing post-training quantization method is difficult to simultaneously consider the difference of different channel activation outliers, the group mode quantization has large dynamic range in the group, and the low bit weight and high bit activation alignment efficiency is low, which leads to the problem of precision decline and inference efficiency limitation, the scheme calibrates the maximum activation amplitude of the input channel by statistical data, divides the significant and non significant activation outlier channels according to the percentile threshold, and respectively uses the logarithmic domain and power function domain smoothing strategy to generate channel smoothing scaling factor, and fuses it into the model parameter to extract the weight multidimensional feature to construct the channel quantization sensitivity index and execute the input channel rearrangement, and further generate the joint quantization scaling factor to realize the integer domain alignment inference. It is suitable for low bit compression deployment and efficient inference scene of large language model in resource limited environment.

The application relates to a stereo matching and disparity map optimization method based on adaptive smooth constraint, which comprises the following steps: acquiring satellite stereo pairs and refined rational polynomial coefficients; processing the satellite stereo pairs and the rational polynomial coefficients by adopting a core line resampling method based on orthographic correction to generate core line image pairs; performing stereo matching on the core line images by adopting a semi-global matching algorithm based on adaptive parameters to generate an initial disparity map; performing optimization processing on the initial disparity map by using an improved weighted least square algorithm to obtain a final disparity map; and generating point clouds by spatial forward intersection based on the final disparity map, performing gridding processing on the point clouds, and acquiring a final digital surface model. Compared with the prior art, the application realizes the balance between noise suppression and detail reservation, thereby improving the accuracy and completeness of the DSM.

This invention discloses a bolt loosening location method and system based on environmental vibration and multi-scale PNN. The location method includes the following steps: S1: Establishing a refined finite element simulation model of the transmission tower and dividing it into multi-scale hierarchical substructures; S2: Simulating bolt loosening conditions in the finite element simulation model and extracting operational modal analysis results; S3: Calculating the frequency change ratio and modal compliance residual matrix norm, and constructing a multi-source damage-sensitive feature vector after weighted fusion; S4: Constructing and training a probabilistic neural network with an adaptive smoothing factor adjustment mechanism; S5: Acquiring on-site environmental vibration data and identifying modal parameters; S6: Extracting and calculating the measured data of the tower using the random subspace identification method, and inputting it into the trained probabilistic neural network to obtain the location results of the main bolt loosening area. This method achieves accurate bolt loosening location through scientific substructure division and multi-source fusion strategies, providing a reference for the safe operation and maintenance of tower bolts.

This application provides a rendering optimization system and method for altered mineral 3D models. The method first divides the 3D model surface into uniform alteration intensity regions and alteration intensity transition zones. Based on the local gradient direction of the alteration intensity transition zone, directional compensation is applied to the boundary diffusion parameter, which characterizes the visual blurring properties of the boundary, to obtain the visual compensation coefficient for the alteration intensity transition zone. Based on the alteration intensity difference between adjacent uniform alteration intensity regions in the 3D model and the inherent width of the alteration intensity transition zone, the visual loss of boundary continuity is determined. The anisotropic smoothing constraint weights are determined through the visual compensation coefficients and the visual loss of boundary continuity when performing color interpolation calculations for each alteration intensity transition zone. Then, adaptive smoothing rendering optimization is performed on the transition zones of the 3D model surface based on all anisotropic smoothing constraint weights. This application's solution can achieve adaptive smoothing rendering of the transition zones on the 3D model surface, thereby improving boundary continuity.

This invention discloses a sparse-view 3D reconstruction method based on block-based geometric alignment and texture decoupling, belonging to the field of computer vision and 3D reconstruction technology. The method includes: acquiring RGB images and prior depth maps from a sparse viewpoint, and initializing a 3D Gaussian sputtering model; eliminating monocular depth local scale drift through block-based local alignment and constructing a second-order normal loss to constrain geometric consistency; achieving adaptive smoothing regularization based on texture-structure decoupling weights guided by normal gradients; generating artifact masks through depth residual analysis, applying soft penalties and performing hard pruning on anomalous Gaussian primitives; jointly optimizing color reprojection loss and geometric, smoothing, and penalty losses, iteratively updating Gaussian parameters, and achieving high-fidelity rendering of the new viewpoint. This invention employs the aforementioned sparse-view 3D reconstruction method based on block-based geometric alignment and texture decoupling, solving the problems of geometric instability, texture-structure coupling interference, and floating artifacts under sparse viewpoints.

This invention relates to the field of image processing technology and discloses an adaptive smoothing and enhancement method for animation capture trajectories based on temporal feature fusion. The method includes the following steps: S1, acquisition and preprocessing of raw trajectory data; S2, extraction of multi-dimensional temporal features; S3, noise and signal identification based on gated feature fusion; S4, generation of adaptive smoothing kernel and trajectory repair; S5, trajectory enhancement and output: post-processing based on physical constraints is performed on the smoothed trajectory to enhance the dynamic rationality of the action, and a smoothed and feature-enhanced animation capture trajectory is output. This invention intelligently distinguishes between real high-frequency motion and noise by fusing kinematic features, spatial correlation features, temporal semantic features, and attention weights, avoiding the over-smoothing distortion phenomenon of traditional methods. Through a multi-scale feature fusion method optimized by mean drift, key motion regions are automatically identified and assigned higher detail preservation weights, achieving differentiated smoothing enhancement.

An apparatus includes an interface and a processor. The interface can be configured to receive pixel data of an environment. The processor can be configured to process the pixel data arranged as a video frame, generate a luminance profile of the video frame in response to a low pass filtering operation, determine a plurality of dehaze intensity weights for the luminance profile, perform adaptive smoothing on each dehaze intensity weight of the plurality of dehaze intensity weights, and generate a dehazed video frame in response to the video frame and the plurality of dehaze intensity weights with adaptive smoothing. Each dehaze intensity weight of the plurality of dehaze intensity weights can correspond to one of a plurality of luminance intervals of the luminance profile. The adaptive smoothing can be configured to prevent bright differences in the dehazed video frame.

The application relates to a kind of beam transverse operating point measurement method and device, method includes: in time sequence to Schottky signal sampling;Current frame power spectrum is obtained based on current frame data;Unified physical frequency axis under current frame target power spectrum is obtained based on current frame power spectrum;Adaptive smoothing filtering is carried out to obtain the smoothed target power spectrum of current frame;The frequency coordinate of the smoothed target power spectrum of current frame is normalized to dimensionless operating point coordinate to obtain the normalized power spectrum of current frame;The normalized power spectrum of current frame is resampled to obtain the resampled power spectrum of current frame;Determine the reference power spectrum of current frame and the reference transverse operating point of current frame;Determine the candidate transverse operating point of current frame;The reference transverse operating point and candidate transverse operating point of each frame in frame sequence are determined by current frame, the preceding several frames of the current frame, and the optimal estimation value of the transverse operating point of the current frame is determined.

The application provides a method and system for improving the diversity of bit-level autoregressive image generation, and relates to the technical fields of artificial intelligence and computer vision, and has the characteristics that in the sampling process of a bit-level autoregressive image generation model, for each generation scale k, the original prediction logits of all token bits at the scale are obtained; based on the original prediction logits, an adaptive temperature coefficient τk is dynamically solved by a binary search algorithm, so that after the logits are scaled by the τk and normalized by softmax, the difference between the actual average maximum bit probability of the current scale k and the preset target average maximum bit probability Sk is less than the preset tolerance; an adaptive temperature coefficient τk is dynamically determined by binary search, so that after the softmax normalization is applied to the τk. The application has the advantages that the generation diversity is improved by adaptively smoothing the early prediction distribution, and the visual quality is maintained by means of energy-based path search; the method mainly operates in the early stage with low computational overhead, and the inference delay is only slightly increased, and the core design can be extended to similar autoregressive generation models.

The application provides a human-machine shared intelligent electric automobile lateral control system and method, relates to the technical field of intelligent automobile control, and comprises the following steps: obtaining a current steering wheel torque and its differential, a longitudinal vehicle speed, a yaw angular velocity and a desired front wheel turning angle; determining a driver intervention intention level according to the absolute value of the torque and determining an initial weight coefficient; constructing a steering quality coefficient by analyzing the time domain fluctuation characteristics of the torque and its differential, and simultaneously correcting the first target front wheel turning angle of automatic driving based on the vehicle speed and the desired turning angle; performing adaptive smoothing filtering on the torque according to the quality coefficient to generate a driver demand turning angle, and then weighting and fusing the first target turning angle and the driver demand turning angle according to the initial weight to obtain a second target front wheel turning angle of shared control and output execution, so that the comfort, smoothness and safety of human-machine shared driving are improved.

The application provides a roadside weak light data synthesis method and device for simulating imaging characteristics of a sensor, and relates to the technical field of computer vision and intelligent transportation, and comprises the following steps: acquiring a daytime roadside image and performing normalization processing; simulating a nonlinear brightness response of the sensor by exposure scaling and Gamma correction; applying a spectral scaling coefficient to simulate spectral shift of a night artificial light source; constructing a dark corner intensity mask to simulate lens optical defects; performing adaptive smoothing on a dark area in a color space to simulate texture loss; establishing a Poisson-Gaussian mixed model to simulate physical noise of the sensor; performing ISP fine tuning and quantization to synthesize a weak light image. In view of the problems that weak light data is scarce in a roadside sensing scene and a traditional image darkening method cannot reflect real physical characteristics, the weak light data generated by the method is consistent with a real night scene in terms of pixel visibility distribution, texture structure reservation and color noise characteristics, and the robustness and detection accuracy of a roadside all-weather traffic sensing algorithm are effectively improved.

The application discloses a fish culture net cage noise tracking method based on optimal smoothing and minimum statistics, and relates to the field of aquaculture. According to the frequency energy distribution characteristics of an input signal, the application automatically adjusts a smoothing parameter, smoothes a power spectrum, and combines minimum value deviation compensation to estimate a smoothing result, so that accurate tracking of a background noise signal is realized, noise power spectrum components can be effectively extracted in a complex noise environment, and additional detection for a target signal is not needed. The application introduces a time-frequency adaptive smoothing parameter mechanism, and fuses minimum statistics to track and estimate a background noise power spectrum. Without training data, the application can effectively determine the noise condition of a sampling signal in a complex noise condition, and solves the problems that existing noise separation methods have poor performance when separating non-stationary noise and need a large amount of training data.

The application discloses a self-supervised heterogeneous knowledge graph learning method based on feature structure homogeneity and long-range heterogeneity. First, a self-expression solver is proposed, which can capture the complementary homogeneity between meta-path and node features, and then obtain homogeneity representation. At the same time, the application designs a path encoder to model various interaction relationships. Through adaptive fusion, the cross-type interaction relationship is explicitly included while reducing noise interference, and long-range heterogeneity is captured. Theoretical analysis verifies that the homogeneity representation has a high-order grouping effect and can effectively capture complementary homogeneity; the path encoder has adaptive smoothing capability to filter noise; and cross-type interaction modeling integrates homogeneity and heterogeneity to introduce more task-related information. Experiments on various data sets including large-scale data sets fully verify the superiority of the method.

The present application relates to the technical field of seismic data processing, and more particularly to a source rupture process inversion method and system based on multi-source observation data; the present application jointly uses multi-source observation data of earthquakes and geodetic survey, combines multi-dimensional advantage resolution capability, and obtains more accurate inversion results; meanwhile, the present application constructs a fault geometry model adaptive to different complexity sources, which is more suitable for the real three-dimensional form of the seismogenic fault; in addition, the present application proposes a space-time scale adaptive smoothing constraint based on the driving of the rupture physical process, realizes the synchronous regulation of the space-time constraint strength by a single hyperparameter, and obviously reduces the difficulty of hyperparameter selection; the present application also realizes the compatibility of at least two fault motion mechanisms by the non-negative constraint of the half-space inequality of the main sliding direction, while prohibiting non-physical reverse sliding, and improves the physical rationality of the inversion model.

The present application relates to the technical field of data processing, and more particularly to a grease oxidation stability prediction method and system based on deep learning, which divides the multi-dimensional monitoring data sequences of different complete grease accelerated oxidation experiments at the same time into a plurality of monitoring data sequence sets; performs data fusion on the monitoring data of all monitoring data sequences in any monitoring data sequence set at the same time to obtain a fused data sequence, and divides the fused data sequence into three stage data sequences; obtains an adaptive smoothing window of each fused data in each stage data sequence, performs smoothing processing on the fused data sequence to obtain a smoothed data sequence; and uses the smoothed data sequences of all monitoring data sequence sets as the input of a CNN network, outputs the predicted value of the grease oxidation stability, effectively suppresses noise and stage differences through phased adaptive smoothing, strengthens the feature expression of the key change segment, and improves the prediction accuracy of the grease oxidation stability.

The application discloses a privacy protection method and system for multi-modal attack behavior adaptive recognition, and relates to the field of artificial intelligence security. The method constructs multi-modal fusion features with time sequence information by quality-aware semantic alignment fusion of multi-modal data; after local differential privacy protection of user behavior features, the multi-modal features are respectively mapped to attack subspaces to generate semantic representations of teacher and student models; in federated training, the knowledge distillation strength is dynamically adjusted based on sample quality and privacy risk signals; according to the modal contribution degree and sensitivity, the gradient is adaptively smoothed and noise injected, and the global model is updated through safe aggregation. The trained model realizes adaptive attack detection and modal-time fine-grained positioning in the attack subspace. The system has the same concept. The application improves the accuracy and explainability of multi-modal attack recognition while providing strict privacy protection.

This invention discloses an intelligent mapping method for power distribution networks based on lattice layout and dynamic weights, relating to the field of power distribution network mapping technology. The method includes constructing a lattice layout system corresponding to the power distribution network. This system uses the geographical coordinates of the area where the power distribution network is located as a reference, and forms several uniformly distributed lattice units through a three-level calculation process involving regional gridded terrain factors to correct equipment density feedback. Specifically, after initial gridding, the side length is corrected by combining terrain factors, and then iteratively adjusted according to equipment density. The adaptive smoothing process is based on GIS vector terrain data, and the boundary is fitted using Bézier curves. Combined with accuracy verification, the method ensures the real-time performance and accuracy of the mapping in complex environments and emergency scenarios. Overall, this method achieves intelligent, high-precision, and efficient dynamic updating of power distribution network mapping, significantly improving distribution network operation and maintenance efficiency, emergency response speed, and fault handling capabilities.

A kind of metal surface defect detection method for sample scarcity, first, different levels of multiscale features are extracted from the input image, and the multiscale features are enhanced and fused to construct a fused feature map;Then the fused feature map is divided into multiple image sub-region feature units, and an adaptive core feature screening strategy is used to construct an image-specific core feature set, and a robust statistical method is used to generate a global prototype feature. Then, an adaptive weight mechanism is introduced to weight the image sub-region features of the image to be detected and reconstruct them, and the abnormal response value of the corresponding image sub-region is calculated according to the reconstruction error to realize abnormal scoring. Finally, the local abnormal response is mapped back to the original image space and optimized by adaptive smoothing, generating an abnormal heat map to realize the positioning and visualization of metal surface defects. The present application has good detection performance and stability under the condition of few samples, and is suitable for metal surface defect detection tasks in industrial scenarios.

The application discloses a kind of based on space-time matching with electric water synchronous metering method and system, and is suitable for agricultural irrigation machine well.The method comprises: collecting electric parameter and water quantity data by hardware interrupt and adding time stamp;Identify the working condition of water pump based on electric parameter characteristic vector;Calculate real-time conversion coefficient;Adaptive smoothing filter update coefficient is used in working condition association;Calculate deviation and compare with dynamic threshold, when exceeding threshold, read historical data and execute three-level classification isolation;Data is packaged and reported.The application realizes millisecond-level synchronous acquisition, working condition adaptive calibration and abnormal classification isolation through acquisition, identification, calibration and abnormal collaborative closed loop, and significantly improves the accuracy of electric water metering.

This invention discloses a method, system, and apparatus for multi-source remote sensing image fusion, relating to the field of remote sensing image processing. The scheme includes acquiring multiple original remote sensing images and performing standard preprocessing to obtain standard remote sensing images; processing these images according to a preset downsampling strategy to obtain degraded remote sensing images at a first spatial resolution; performing adaptive robust smoothing processing on these images using a temporal robustness processing strategy to obtain robust remote sensing images corresponding to each time point and at the first spatial resolution; and performing upsampling processing on these images according to a preset fusion strategy to obtain a fused remote sensing image at a second spatial resolution. This scheme ensures geometric stability and spectral physical consistency through standard preprocessing, and reliably achieves robust processing, adaptive smoothing, and information preservation of degraded remote sensing images through a temporal robustness processing strategy, improving the continuity and consistency of robust remote sensing images over long time scales, and ultimately determining the fused remote sensing image, which is beneficial for anomaly monitoring.

The application discloses a lithium battery charge prediction method based on adaptive smooth attention and local time fusion, and belongs to the technical field of power battery state monitoring and intelligent prediction, and comprises the following steps: collecting and pre-processing battery operation data to obtain an input feature matrix; an encoder embedded with an adaptive smooth attention mechanism is used to decompose an input query matrix into a trend component and a residual query component, the trend component and the residual query component are fused after local autoregression and global attention processing, and encoded features are output; a decoder is used to generate preliminary prediction features; an anti-noise feature sequence is obtained through gated residual fusion of a local time fusion architecture; time continuity is constrained through a long short-term memory network to obtain smooth time sequence features; and finally, a state of charge prediction value is mapped and output, and a warning is given. The application adopts the above method, improves the accuracy and robustness of state of charge prediction, maintains high stability under a wide temperature range and complex dynamic working conditions, effectively suppresses noise interference, and realizes active warning.

This invention discloses a method and system for improving the success rate of wide-lane ambiguity fixation in BeiDou dual-frequency positioning, belonging to the field of high-precision positioning technology. It solves the problems of existing methods being unable to adapt to dynamic changes in observation quality, suffering from old information contamination, and difficulties in setting the smoothing window length. The method includes: constructing a MW combined observation model for BeiDou dual-frequency positioning; establishing a time-decay-based weight allocation function to recursively smooth the wide-lane ambiguity floating-point solution sequence; performing integer fixing and reliability verification based on the smoothed wide-lane ambiguity floating-point solution estimate; and dynamically adjusting the exponential weighting factor according to the measurement residual to achieve adaptive smoothing. This invention introduces a time-decaying weight function to give higher weight to recent observations and establishes a recursive update model. Simultaneously, it introduces a residual-driven dynamic decay factor adjustment mechanism, effectively improving the reliability and timeliness of wide-lane ambiguity fixation in BeiDou dual-frequency positioning.

The application discloses a kind of remote sensing vegetation index time series missing value reconstruction methods of adaptive smoothing and trend constraint coordination, comprising: obtaining the remote sensing vegetation index time series data of pre-processing, extract effective observation value and identify missing value to construct one-dimensional time series;The observation density of time series is calculated, and the smoothing parameter and trend fusion weight are determined accordingly;Based on smoothing parameter, an adaptive weighted Whittaker smoothing model is constructed to obtain a preliminary smoothing sequence by fitting;According to the number of effective observation value, the trend reference value at the missing value is calculated by adaptively selecting the trend estimation method;Based on the trend fusion weight, the preliminary smoothing sequence and the trend reference value are weighted and fused to generate the filling value, and only the missing value is replaced to obtain the preliminary reconstruction sequence;Multi-level physical constraint verification is applied to the filling value, and the final reconstruction result is output.The application overcomes the defects of excessive smoothing and abnormal interpolation of traditional methods, and realizes the unity of data authenticity, continuity and physical rationality.

The application discloses a single information mixed differential privacy data distribution estimation method and system and electronic equipment: receiving multiple pieces of privacy data input by multiple users, and mapping the multiple pieces of privacy data into multiple pieces of perturbation data according to a specific probability distribution; then performing a uniform random permutation operation on the multiple pieces of perturbation data to obtain multiple pieces of mixed perturbation data; and then performing distribution estimation on the multiple pieces of mixed perturbation data based on an adaptive smoothing expectation maximization aggregation algorithm to obtain an estimation result. The application is a single data mechanism, that is, each user only needs to send one piece of data, thereby reducing communication cost. The application is used for numerical data distribution estimation, perturbation is designed through an optimal perturbation parameter, the perturbation result can maximize the retention of real privacy data information input by the user, and the estimation result is more accurate. The algorithm of the application introduces the adaptive smoothing expectation maximization aggregation algorithm, and the security and accuracy of estimation can be realized while the robustness is enhanced.