45 results about "Spatially adaptive" patented technology

The invention discloses an echocardiogram myocardial motion estimation method and system based on a structure perception diffusion model, and relates to the technical field of ultrasonic video data processing. The invention aims to solve the problem of myocardial motion estimation in an ultrasonic video. According to the technical key points, the method comprises the following steps: (1) preprocessing collected echocardiogram video data; (2) establishing an echocardiogram myocardial motion estimation model; (3) constructing a diffusion model motion estimation module of structure perception; (4) constructing a spatial adaptive up-sampling module; (5) training the network by using the existing data to obtain a network model; and (6) estimating the myocardial movement of the echocardiogram video by using the trained model. According to the method, the cardiac structure information and the multi-scale correlation contained in the echocardiogram can be fully utilized to automatically estimate the cardiac muscle movement of the echocardiogram, and the accuracy and the robustness of the algorithm are effectively improved.

The invention discloses a data management system based on unmanned aerial vehicle surveying and mapping, and particularly relates to the technical field of data management. The method comprises the steps of collecting unmanned aerial vehicle surveying and mapping data, and performing space coordinate registration and unified coding; performing spatial region division according to the terrain elevation variable coefficient and the obstacle projection coverage rate to obtain a high-density verification region and a low-density verification region; for the high-density verification area, calculating the feature matching degree of an overlapping area of adjacent frames through a local feature matching algorithm, and outputting first correction data; performing data compensation on the low-density verification region by adopting spatial interpolation and cubic spline interpolation to generate second correction data; generating a full-scene reconstruction data frame based on the first correction data and the second correction data; based on full-scene data frame reconstruction, region weight factor distribution, differential error correction coding execution and unmanned aerial vehicle surveying and mapping result data output, the spatial adaptability of data verification is effectively improved, and the overall accuracy and integrity of unmanned aerial vehicle surveying and mapping data are improved.

The invention relates to the technical field of medical image processing, in particular to a spatial self-adaption and thermodynamic diagram guided retina image segmentation method, which comprises the following steps of: acquiring retina image data containing three-dimensional information and at least one focus, optimizing an existing SAM model, and constructing a segmentation model, the architecture of which comprises the steps of taking a visual Transform as a backbone and taking a thermodynamic diagram as a template; designing and introducing an image encoder of a spatial adaptive feature enhanced Transform block, a dual-stage mask decoder and a thermodynamic diagram guided mask self-attention module, adopting a training process of a joint loss function supervision model based on cross entropy loss and Dice loss, and inputting an image to be segmented into the model to obtain a segmentation result; the scheme has the advantages that the three-dimensional structure information of the retina image can be fully mined, the segmentation precision of different types of lesions is remarkably improved, the problem of class imbalance is relieved, and high-precision and high-efficiency segmentation of the retina lesions is realized.

The invention relates to the technical field of sensing navigation, in particular to a patrol robot based on sensing obstacle recognition, which comprises a path dynamic priority module, a high-flow avoidance module, a bedside dynamic area module, an approaching risk monitoring module and a path coverage state module. According to the method, node pressure distribution is adjusted and optimized through dynamic priorities, path flexibility is improved, a self-adaptive avoidance area is defined synchronously based on a bed displacement trend and a safety redundancy range, the path congestion risk is reduced, direction deviation is controllable, detour redundancy is reduced, the dynamic avoidance area shrinks or expands along with environment changes, and space adaptability is enhanced. The approach risk level is combined with the angle change rate and the distance decay rate to quantify the potential risk, the coverage overlapping proportion threshold linkage turning freezing mechanism prevents path conflicts, the multi-dimensional decision basis improves the obstacle avoidance response speed and safety, and compared with a static rule, the adaptability to ICU high-frequency environment changes is enhanced, and the method has the advantages of being simple in structure and convenient to operate. A pressure redistribution and risk grading mechanism reduces obstacle avoidance delay.

The application relates to the field of image data processing and discloses a texture preservation type image denoising and enhancing method based on a generative adversarial network, which comprises the following steps: collecting an original image signal, calculating low-frequency subband data and high-frequency subband data by using discrete wavelet transform; calculating the gradient amplitude of the low-frequency subband data to generate a structure saliency gradient map; establishing an inverse mapping relationship based on the structure saliency gradient map to generate a spatially adaptive dynamic gating threshold; performing statistical gating on the high-frequency subband data by using the dynamic gating threshold to generate a high-pass gain coefficient and a low-pass suppression coefficient; and weighting and filtering the high-frequency subband data by using the coefficient and reconstructing a target output image. The application solves the problem that weak texture signals are prone to be lost under non-uniform illumination by establishing a cross-band modulation logic of structure flow to texture flow, and realizes non-structured noise filtering and structured micro-texture recovery without relying on semantic labels.

The invention discloses a vehicle re-identification method and device, equipment and a medium, relates to the technical field of computer vision, and aims to solve the problem that attribute information of a vehicle cannot be utilized due to lack of data annotation by extracting a semantic feature vector related to the color of the vehicle and a semantic feature vector related to the type of the vehicle. The vehicle color features and the vehicle type features are aligned to the spatial dimension of the vehicle global features, the two aligned features are spliced into attribute features, and the attribute features pass through a convolutional layer and a sigmoid activation function to generate a spatial adaptive color space attention map and a spatial adaptive type space attention map respectively; in the process, the weight of the vehicle color feature and the weight of the vehicle type feature are adaptively and dynamically allocated to improve the feature representation capability, and finally, the global feature capable of representing the appearance feature and the visual feature is spliced with the two weighted features after weight adjustment to perform vehicle re-identification, so that the accuracy of vehicle re-identification is improved.

This invention relates to the field of three-dimensional cloud field forecasting technology, and specifically to a spatially scalable forecasting method, system, device, and medium for three-dimensional cloud fields. The method includes: constructing a first basic training set based on preprocessed first historical radar network data; inputting the first basic training set into a model to obtain an initial forecasting model; inputting various specialized historical data into the initial forecasting model to train specialized identification capabilities, obtaining a specialized enhancement model; constructing a second basic training set based on preprocessed second historical radar network data, and processing the second basic training set based on the specialized enhancement model to select and construct multiple specialized training sets; inputting the second basic training set and the multiple specialized training sets into the specialized enhancement model for multi-resolution joint training to obtain a target forecasting model. This method solves the problems of insufficient spatial adaptability and specialized identification capabilities of three-dimensional cloud field forecasting models under complex weather conditions.

Belonging to the field of computer vision technology, this invention provides a method and apparatus for teacher-assisted atomic motion recognition based on an improved end-to-end network. [Solution] The method determines time-dimensional and spatial-dimensional video features from the target teacher's current lesson video based on an improved target end-to-end network, generates multi-window feature groups, merges the multi-window feature groups using a frame selection network, and recognizes the target teacher's atomic behavior based on the fusion result. This method embeds spatially adaptive units and time-series adaptive units within the end-to-end network, performs feature extraction on each, and reduces video memory consumption by training only the parameters of these units during network training, thereby effectively improving the accuracy and efficiency of atomic behavior recognition, reducing memory requirements, and lowering deployment difficulty.

The application provides a compression method based on information density driving and adaptive quadtree partitioning, comprising the following steps: S1, obtaining an original high-dimensional visual token set and screening to obtain a visual token sub-set meeting the requirements; S2, using a spatially adaptive recursive quadtree partitioning mechanism, reconstructing the visual token sub-set into a token grid according to the two-dimensional spatial position, and partitioning an information-rich region and a low semantic density region, performing fine-grained partitioning and token retention strategy on the information-rich region to obtain a first compressed token sequence, and performing token merging and down-sampling strategy on the low semantic density region to obtain a second compressed token sequence; S3, splicing the first compressed token sequence, the second compressed token sequence and the text token sequence to obtain a final token sequence, which is input into a multi-modal large model for cross-modal reasoning. The beneficial effect is that the application can maintain the understanding performance and spatial structure integrity of multi-modal generalization under extremely low token budget.

The invention relates to the technical field of three-dimensional cloud field forecasting, in particular to a three-dimensional cloud field space expandability forecasting method, system and device and a medium. The method comprises the following steps: constructing a first basic training set based on preprocessed first historical radar networking data; inputting the first basic training set into the model to obtain an initial forecasting model; inputting various special historical data into the initial forecasting model for special identification ability training to obtain a special improvement model; constructing a second basic training set based on the preprocessed second historical radar networking data, processing the second basic training set based on a special lifting model, and screening and constructing a plurality of special training sets; and inputting the second basic training set and the plurality of special training sets into a special lifting model for multi-resolution joint training to obtain a target forecasting model. The problem that a three-dimensional cloud field forecasting model is insufficient in space adaptability and special identification capability under complex weather conditions is solved.

This invention discloses a data management system based on UAV mapping, specifically relating to the field of data management technology. It involves collecting UAV mapping data and performing spatial coordinate registration and unified encoding; dividing spatial regions based on terrain elevation variation coefficient and obstacle projection coverage to obtain high-density and low-density verification zones; calculating the feature matching degree of overlapping areas in adjacent frames using a local feature matching algorithm for the high-density verification zone, and outputting first correction data; performing data compensation using spatial interpolation and cubic spline interpolation for the low-density verification zone to generate second correction data; generating a full-scene reconstruction data frame based on the first and second correction data; allocating regional weight factors based on the full-scene reconstruction data frame, performing differential error correction encoding, and outputting UAV mapping results data, effectively improving the spatial adaptability of data verification and enhancing the overall accuracy and completeness of UAV mapping data.

The present application relates to the technical field of medical image processing, in particular to a kind of space adaptive and heat map guided retinal image segmentation method, comprising: obtaining the retinal image data containing three-dimensional information and at least one focus, optimizing existing SAM model, construct a segmentation model, its architecture contains with visual Transformer as backbone, and design is introduced space adaptive feature enhancement Transformer block image encoder, two-stage mask decoder and heat map guided mask self-attention module, using the combined loss function based on cross-entropy loss and Dice loss composition supervises the training process of model, the image to be segmented is input into model to obtain segmentation result;The advantages of the scheme are that the three-dimensional structure information of retinal image can be fully tapped, the segmentation accuracy of different types of lesions can be significantly improved, and the problem of class imbalance can be alleviated, and high-precision and efficient segmentation of retinal lesions is realized.

The invention provides a self-adaptive phase field sea ice image segmentation method. The method comprises the following steps: S1, initializing a phase field function based on the intensity information of an input sea ice image; s2, calculating an edge feature map of the input sea ice image, and extracting multi-scale texture features of the input sea ice image to form texture feature items; s3, determining a spatially adaptive evolution speed parameter and a spatially adaptive coupling parameter according to the edge feature map; s4, iteratively updating the phase-field function based on the evolution equation according to the evolution speed parameter of the space self-adaption and the coupling parameter of the space self-adaption; and S5, after the phase-field function reaches a convergence condition, obtaining a final phase-field function, and performing threshold processing on the final phase-field function to generate a binarized sea ice segmentation result image. According to the method, accurate and robust segmentation of the complex sea ice scene is realized by fusing the low-level features of the image and the internal constraints of the physical model, and the method has good generalization ability and relatively low sample dependence.

This invention discloses a multi-branch parallel weighted and multi-scale spatially adaptive infrared image processing method and system. It reconstructs the output data of an infrared detector into a two-dimensional grayscale image matrix; obtains three branch output images through parallel processing of three-branch complementary feature extraction operators; obtains a base image through pixel-level weighted fusion using local spatial information entropy mapping; normalizes and inverts the base image to obtain an inverted image; and processes the inverted image based on a spatially adaptive exponential enhancement model using a multi-scale Gaussian ratio factor to obtain the final enhanced image. This invention not only achieves heterogeneous branch complementary fusion with adaptive parameter determination, enabling the extraction of subtle target details, and low computational complexity, but also possesses potential for dataset generation and migration to embedded real-time systems. Furthermore, it addresses the challenge of existing infrared image enhancement techniques in balancing noise suppression, target detail preservation, and real-time processing.

This invention relates to the field of semiconductor technology, specifically to a method and system for online control of in-plane electrical performance uniformity on a wafer. The method includes: acquiring key process parameters at multiple locations within the wafer plane online during wafer fabrication; inputting these parameters into a pre-trained machine learning model, which models the spatial correlation between locations and predicts the estimated electrical performance at different locations within the wafer plane; subsequently, generating a spatially adaptive process recipe for differentiated process compensation in different regions of the wafer based on the difference between the estimated and target performance; and finally, adjusting at least one process parameter in subsequent process steps according to this recipe to proactively correct electrical performance deviations. This invention achieves a shift from offline feedback to online feedforward, and by integrating spatial modeling and intelligent decision-making, it can suppress wafer-level process fluctuations in real time, proactively, and accurately, significantly improving chip electrical performance uniformity and manufacturing yield.

A method of reconstructing a hyperspectral data cube from an encoded and / or spectrally sheared image and an unencoded and spectrally unsheared image of a scene comprises providing the encoded and / or spectrally sheared image and an unencoded and spectrally unsheared image to a neural network comprising an encoder-decoder structure having an encoder portion and a decoder portion; performing one or more primary convolution operations using the encoder portion; and performing one or more transpose convolution operations using the decoder portion; performing spatially adaptive normalization and / or inverted attention normalization on an output of primary convolution operations and / or transpose convolution operations wherein the spatially adaptive normalization and / or inverted attention normalization is performed using a representation of an encoding scheme used to encode the at least one encoded and / or spectrally sheared image and an unencoded and spectrally unsheared image of the scene.

The application relates to the technical field of data processing, and discloses a forestry carbon sink satellite remote sensing data correction method. The method is based on a satellite transit time window to implement multi-source time synchronization collection on satellite-borne, airborne and ground measured data, combines micro-meteorological data, digital elevation model data and crown laser radar data to simulate carbon sinks at different elevations at a single time point, and completes simulation correction through remote sensing radiation transmission coupling and eddy correlation data assimilation; further, according to the deviation of measured values and simulated values at different elevations, elevation-specific correction parameters are calculated, a three-dimensional correction parameter field of space-time-elevation is constructed and is adaptively updated; and finally, the correction parameters are used to uniformly correct, weightedly fuse and optimize residuals of satellite-borne data and airborne data. The application can improve the correction accuracy, spatial adaptability and processing efficiency of multi-source forestry carbon sink detection data under complex terrain conditions.

The invention discloses a three-view multi-resolution three-dimensional sensing method based on distance sensing, and relates to the field of automatic driving three-dimensional environment sensing. According to the method, a near-field high-resolution and far-field low-resolution double-branch architecture is constructed, and cross-resolution initialization and a multi-scale feature fusion mechanism are combined, so that high-precision perception of a short-distance target and efficient modeling of a long-distance target are realized. In the reasoning stage, a spatial adaptive decoding strategy is adopted, high-resolution and low-resolution features are fused in a short-distance region, and low-resolution features are directly adopted in a long-distance region, so that resolution adaptive three-dimensional perception is realized. Compared with a traditional single high-resolution method, the method has the advantages that the perception precision is kept, meanwhile, the video memory occupation and the calculation overhead are remarkably reduced, the memory consumption of an experimental result table is remarkably reduced by about 26.3%, the reasoning speed is increased to the quasi-real-time level, and the method has the application value of actual deployment on a vehicle-mounted platform.

This invention provides a method and system for predicting ablation temperature fields with full-scale recalibration and spatial compensation, relating to the interdisciplinary fields of biomedical engineering, medical imaging computing, and artificial intelligence. The method involves acquiring a three-dimensional tumor mask, a three-dimensional needle path mask, and an initial specific absorptivity field of the case to be predicted, and stitching them together to obtain a multi-channel input feature map. Feature extraction is performed on the multi-channel input feature map, and semantic feature extraction is performed on the obtained multi-scale encoded feature map to obtain a high-level semantic feature map. Parallel feature extraction is performed using dual-path branching to obtain a first feature map and a second feature map, respectively. Spatially adaptive weighted fusion processing is performed using a gated decision block to obtain a three-dimensional ablation temperature field prediction map. This invention solves the problems of insufficient prediction accuracy, loss of local high-temperature details, poor generalization across cases, and sub-voxel-level spatial positioning bias caused by multi-level sampling in existing microwave ablation temperature field prediction technologies.

A method, device, electronic equipment, and storage medium for controlling in-vehicle air quality are disclosed, relating to the field of vehicle environmental control technology. The method includes: detecting air quality data through an oxygen generation module, achieving multi-sensor fusion to ensure the diversity of air quality data; simultaneously, oxygen generation and purification functions can be used independently or quickly combined, improving equipment utilization and spatial adaptability; furthermore, determining the air quality state inside the vehicle cabin based on the acquired air quality data and multiple thresholds, and then switching the operating states of the oxygen generation module and air purification module according to different air quality states; achieving dynamic adjustment of operating states through a unified control algorithm to reduce energy consumption and improve the efficiency of in-vehicle environmental control.

A system and method are disclosed to address the need for adaptive exposure in high dynamic range (HDR) images. The solution can leverage recent advances in the use of virtual reality (VR) headsets and augmented reality (AR) displays equipped with infrared (IR) eye-tracking devices. The gaze vector determined by the eye-tracking device identifies one or more fixation points on the image corresponding to areas where underexposure exists. Exposure around the fixation points can be adaptively corrected using image processing techniques. Using spatially adaptive exposure, the resulting image, a kind of foveal image, can be rendered with sufficient detail on a low dynamic range (LDR) display.

The invention discloses an intelligent surveying and mapping analysis platform for multi-source geographic information data fusion, and relates to the technical field of surveying and mapping analysis. Calculating the distribution probability based on a sliding window and quantifying the uncertainty; standardizing the data and then constructing a joint histogram, and calculating a cross-data-source dependency relationship; constructing a convex optimization model containing local entropy and global mutual information, and iteratively solving a spatial adaptive weight; combining a weight matrix and mean filtering to eliminate fusion mutation; non-local mean filtering is adopted to reinforce the edge, and a radial basis function is utilized to fill a data missing area. According to the method, the convex optimization model containing the local entropy constraint and the global mutual information constraint is constructed, so that the adaptive compensation of the spatial heterogeneity of the multi-source data is realized, and a more accurate fusion result is realized in a complex geographic scene.

The application discloses a kind of space adaptive fluid simulation method based on smooth particle hydrodynamics, comprising: the three-dimensional model of boundary solid object is converted into SDF model;In simulation scene, import fluid particle set, SDF model and its gradient field;In single simulation time step, the scale function value of particle is calculated, while applying wake protection mechanism, the decay of scale function value of fine particle near boundary solid is delayed;The best scale of particle is calculated, whereby the particle is classified, and the scale of particle is adjusted according to the category of particle;The dynamics calculation of fluid particle is carried out, while processing the coupling of fluid particle and boundary solid, using time mixing scheme to stabilize numerical field, update the physical properties of particle and the physical properties of boundary solid.The scheme of the application can more effectively refine the coupling details of fluid and multiple dynamic boundary solid objects while improving the simulation efficiency, and effectively capture the complex dynamic behavior below fluid surface.

This invention discloses a remote sensing land cover data fusion method based on category-accuracy geographic weighting, comprising: preprocessing multi-source land cover products to unify spatial reference and classification systems; constructing a validation sample set using stratified random sampling combined with visual interpretation; providing two methods for determining the number of neighborhood samples N, with an adaptive method dynamically adjusting N based on sample density and product category consistency, and adjusting the neighborhood radius to adapt to the sample distribution; calculating pixel scales GWUA and GWPA based on spatial weights, calculating the fusion probability using GWUA, and arbitrating the average GWPA when balancing. This invention achieves spatially adaptive estimation of accuracy, improving the accuracy and stability of the fusion results, and providing reliable data support for ecological monitoring and land management.

An image processor for transforming an input image, the image processor being configured to implement a trained artificial intelligence model, wherein the image processor is configured to: receive the input image; process the image through the trained artificial intelligence model based on (i) content of the input image and / or (ii) features extracted from the input image to: (i) determine a set of image filters; (ii) for each of a plurality of sub-regions of the image, select an image filter from the set of image filters; for each of a plurality of sub-regions of the image, apply the respective image filter to the sub-region or to features extracted from the sub-region. This can enable selection of filters from a discrete set of learnable and decorrelated filters in a differentiable manner to enable content-based spatial adaptation.

Spatially adaptive quantization-aware deblocking filters are disclosed. The spatially adaptive quantization-aware deblocking filters are used to encode or decode video or image frames. The deblocking filters receive a reconstructed frame generated based on dequantized and inverse transformed coefficients of the video or image frames. The reconstructed frame is filtered according to adaptive quantization field data of the video or image frames. The adaptive quantization field data represents weights applied to quantization values used at different regions of the video or image frames. The number of blocking artifacts remaining in the generated filtered frame is determined. The adaptive quantization field data is then adjusted based on the number of blocking artifacts. The filtered frame is then filtered according to the adjusted adaptive quantization field data. The generated re-filtered frame is then output to an output source, such as for transmission, display, storage, or further processing.

Disclosed are systems and methods for rendering personal sound zones (PSZs) with head tracking using multiple loudspeakers, utilizing a Spatially Adaptive Neural Network (SANN) to dynamically generate and update audio filters that create acoustically bright and dark zones based on the listener head positions. First, the SANN encodes listener head coordinates and outputs corresponding PSZ filters as either time-domain impulse responses or frequency domain complex -valued coefficients. Then, acoustic transfer functions (ATFs) corresponding to the rendering loudspeakers and control points within PSZs are simulated at random room acoustics conditions and optionally measured in a specific rendering environment. The combined ATFs along with a unified loss function that incorporates multiple PSZ design objectives, are utilized in training the SANN for randomly sampled listener positions within the rendering area. Finally, the trained SANN is used for generating PSZ filters in real-time given a stream of head position data from a head tracking device.

This application provides a geometric-aware compression and reconstruction method and system for 360-degree panoramic images. The method includes generating a position code based on the latitude and longitude coordinates of the equidistant cylindrical projection of the 360-degree panoramic image, serving as geometric prior information; using the position code to drive a geometrically aware deformable convolutional network in the encoding end to perform geometrically adaptive feature extraction, obtaining coded features; in a hierarchical prior network, fusing the position code and the coded features to perform spatially adaptive probability distribution estimation, and generating a binary bitstream through an arithmetic encoder; decoding the binary bitstream through an arithmetic decoder to recover latent space features; and at the decoding and reconstruction end, based on the position code, performing multi-stage correction on the latent space features through a reconstruction-aware spatial attention module to restore high-frequency details and output a reconstructed image. This application can achieve the goal of improving the compression efficiency and reconstruction quality of panoramic images.

A kind of hydraulic jack comprehensive test platform, including pedestal, the pedestal is equipped with lifting frame and test seat, test seat is equipped with self-clamping station, lifting frame is equipped with processing structure, the bottom of processing structure is equipped with test inductor matched with test seat, test inductor is connected with processing structure by data line, the bottom of pedestal is equipped with base structure.Using the above structure, the overall structure of the device is more stable, the space adaptability is strong, the test is more accurate;Self-adapting different size hydraulic jack placement test, clamping limit, stable cooperation during measurement;Adaptive height adjustment, in different stroke range test of hydraulic jack, test data is more comprehensive, measurement analysis is accurate in place.