76 results about "Artifact suppression" patented technology

The invention relates to the technical field of medical image processing, in particular to an artificial intelligence assisted intraoperative imaging method, which comprises the following steps: S1, preprocessing a multi-modal medical image, segmenting and recognizing an anatomical structure by a deep learning model according to the preprocessed image, measuring anatomical parameters based on a segmentation and recognition result, and generating an operation planning path by artificial intelligence according to the anatomical parameters; s2, collecting a C-shaped arm perspective image stream in real time, dynamically tracking space coordinates of a surgical instrument, comparing the position of the instrument with a surgical planned path, calculating offset, and when the offset is greater than an offset threshold, outputting correction guidance through an AR superposition layer; and S3, monitoring an image quality index in real time, dynamically adjusting exposure parameters through a reinforcement learning model, and when a metal implant is detected, switching a dual-energy-spectrum mode and executing an artifact suppression algorithm. According to the method, preoperative precise planning and intraoperative assistance are realized through artificial intelligence, the problems of poor image quality and high radiation risk are solved through technical optimization, and the method has important clinical application value.

The invention discloses a rapid imaging method and system for a multi-time-sequence magnetic resonance image group, particularly relates to the technical field of multi-time-sequence magnetic resonance image processing, and is used for solving the problems of low imaging speed and poor time sequence consistency. According to the method, sampling reference distribution is constructed in a frequency domain, the reconstruction difficulty is predicted by using graph convolution, a multi-scale under-sampling mask is generated, and sampling is distributed according to channel sensitivity, so that high information priority and sparse controllability are realized; sending the registered sparse data into a space-time-parameter three-domain network, extracting structural, dynamic and parameter characteristics, and obtaining an image group with a clear structure and a stable time sequence through cross-domain fusion, non-rigid body registration and wavelet domain artifact suppression; and forming a comprehensive score by using four indexes output by the quality scoring network, performing reinforcement learning feedback, adaptive adjustment of a sampling rate, a mask, a reconstruction parameter and closed-loop optimization, and finally outputting the comprehensive score together with a sequence parameter, a timestamp and a score, and generating a structure comparison diagram and a time curve to facilitate quality control and presentation.

The invention discloses a brain power supply imaging method and system based on a physical information neural network. The method comprises the following steps: establishing a three-layer boundary element method head model based on a Maxwell equation set under quasi-static approximation and calculating a lead matrix; carrying out filtering, artifact suppression and standardization processing on the multi-channel electroencephalogram signals; inputting the standardized signal into a neural network encoder to extract a time sequence feature tensor, estimating the source current density based on the feature tensor, and performing forward calculation by using the lead matrix to obtain a predicted scalp potential; constructing a mixed loss function consisting of a data fidelity item, a physical consistency item and a physical heuristic regularization item, and performing network training and optimization by taking minimization of the loss function as a target; and when the training meets a convergence condition, outputting a visualization result of the brain endogenous current density spatial distribution. According to the method, electromagnetic physical constraints are introduced, so that the physical consistency, interpretability and stability of electroencephalogram inversion are improved.

The invention discloses an SPR (Surface Plasmon Resonance) image optimization processing method based on image segmentation and edge enhancement, which comprises the following steps: acquiring SPR image data, and preprocessing to generate standardized SPR image data; inputting a structure boundary extraction model constructed based on CGNet, generating a structure boundary label graph, and aligning the structure boundary label graph with the image; gradient amplitude and local entropy mutation detection is executed, and an artifact guide graph is generated; respectively inputting the standardized image into details and context branches of the improved CSDNet, and extracting edge and semantic feature maps; inputting a guide perception gating module, executing structure enhancement and artifact suppression fusion, and generating a fusion feature map; inputting into a multi-scale detail recovery module, and outputting an edge enhanced image; and executing structural similarity and marginal definition scoring based on the original image and the enhanced image, and generating an optimization result. According to the method, synchronous optimization of SPR image edge enhancement and structure maintenance is realized, and the image definition and diagnosis availability are remarkably improved.

The invention provides a brain-computer interface decoding method and system based on multi-modal signal fusion, and relates to the technical field of brain-computer interfaces. The method comprises the following steps: S1, acquiring a multi-mode signal, unifying a time reference and starting a sliding window detection intention; s2, de-prompting, sequential bias checking and permutation correction, artifact suppression and band-pass filtering are carried out; s3, channel and modal quality is calculated, and a reliability weight and a rejection list are generated and marked; s4, multi-scale coding is carried out, cross-modal alignment is realized through learnable time delay compensation, and a first fusion representation is obtained; s5, performing layering and cross attention fusion under reliability gating so as to align and reconstruct constraint to keep semantic consistency, and obtaining joint semantic representation; and S6, sequence-level decoding is carried out to output semantics / states / instructions. According to the method, bias and noise are suppressed, time delay difference and cross-main-body equipment change are resisted, and decoding accuracy, robustness and practicability are remarkably improved.

The invention relates to the field of medical image processing, and discloses a multi-mode magnetic resonance aneurysm occlusion assessment method and system. The method comprises the following steps: acquiring scanning configuration and a multi-modal magnetic resonance sequence, and generating fused image data through cross-modal registration, artifact suppression and denoising processing; image segmentation, three-dimensional structure reconstruction and topology correction are carried out, and a three-dimensional anatomy and blood flow model is obtained; then, through gridding, index normalization and time sequence alignment processing, individual feature vectors are extracted; and finally, obtaining a blocking mechanism score and probability based on multi-model fusion reasoning, completing three-dimensional visual rendering, and backfilling key parameters to a configuration structure to form closed-loop optimization. According to the method, multi-modal data are effectively integrated, the evaluation accuracy is improved, and self-adaptive optimization and result traceability of the system are realized through parameter backfilling.

The invention discloses a real-time self-adaptive electroencephalogram artifact suppression and enhancement embedded system. The system aims at overcoming the limitation of an existing portable EEG device in the aspect of artifact processing, various artifacts in EEG signals are recognized and removed in a low-power-consumption, high-precision and real-time mode, and meanwhile the quality of effective EEG signals is optimized and enhanced. One key innovation of the invention lies in breaking through the limitation of single EEG signal processing, and accurate separation and correction of artifacts are carried out through the assistance of EOG and EMG. For example, the EOG signal is used for assisting eye movement artifact removal through synchronous correlation analysis and an adaptive regression filter; the EMG signal is used for assisting myoelectricity artifact correction, and suppression is carried out through spectral analysis and dynamic adjustment of filter parameters. In addition, the system also has the functions of real-time detection and repair of electrode artifacts. According to the system, high-quality EEG signals close to offline processing are provided on portable embedded equipment with limited resources, and the system has universality and high-precision application potential and can be widely applied to a plurality of frontier fields such as medical diagnosis, brain-computer interface (BCI), cognitive ability enhancement, neurological rehabilitation training, sleep disorder monitoring and human-computer interaction.

The invention discloses a multi-exposure fusion HDR image reconstruction method based on a flexible lens, and the method comprises the steps: carrying out the image reconstruction of an LDR image sequence through a trained exposure image reconstruction model, and obtaining an HDR reconstruction image; the exposure image reconstruction model comprises a pseudo twin space-frequency domain attention module, an exposure guide fusion module and an exposure recovery module; after features of an input sequence are extracted through an encoder, feature modulation is carried out through a pseudo twin space-frequency domain attention module, weighted fusion is carried out on a modulated feature map through an exposure guide fusion module, illumination recovery is carried out through an exposure recovery module, shallow features of a reference image are fused, and finally an HDR reconstruction image is generated through a decoder. According to the method, the limitation of a traditional model on the fixed number of lenses is broken through, self-adaptive processing on the input sequence of any length is realized, and meanwhile, the detail retention and artifact suppression capabilities are remarkably improved through structural tensor loss.

A plant three-dimensional reconstruction method for a weak texture image and scale distortion comprises the following steps: step 1, recovering an initial Gaussian point cloud and camera parameters through an SfM method by using a multi-view RGB image; 2, differential rendering is executed in training iteration, a predicted image is compared with a real image, and a loss function containing various constraints is calculated to serve as a basis for gradient solving and parameter updating; 3, calculating a pixel weighted average gradient and an artifact suppression coefficient of each Gaussian in a visible view angle; 4, updating Gaussian parameters through back propagation, and adaptively triggering Gaussian cloning, splitting or deleting operation based on the pixel gradient, the covariance matrix and transparency; and 5, dynamically adjusting the Gaussian projection matrix according to the focal length and the resolution of the camera during rendering so as to maintain scale consistency, and performing anti-aliasing rendering in combination with pixel-level super-sampling. According to the method, the reconstruction precision and the structure reduction capability of the virtual plant in the weak texture region are remarkably improved.

The invention belongs to the technical field of medical equipment, and provides a tumor radiotherapy system based on multi-modal image and breathing waveform fusion, which comprises a data acquisition and processing module, a tumor position prediction module and a radiotherapy determination module, based on an elastic registration method, through feature space alignment, the space-time registration problem of a multi-source image is effectively solved, particularly artifact suppression and small tumor recognition of a rib overlapping region are innovatively optimized, and the target region sketching precision in a complex anatomical environment is remarkably improved; by means of a two-way convolution module, a prediction model with respiratory phase sensing capacity is constructed by analyzing space-time correlation between respiratory signals and image features, tumor motion trail tracking is achieved, tumor centroid coordinates are predicted, and the technical limitation of traditional single-mode tracking is broken through.

The invention provides a brain-computer interface interaction method and system based on a neural oscillation mode, and relates to the technical field of brain-computer interface interaction. Clean data streams are obtained through multi-lead electroencephalogram collection and filtering artifact suppression, multi-frequency oscillation features are extracted and fused, an end-to-end time delay model is established to achieve phase compensation in combination with individualized dominant frequency and incubation period dynamic calibration, candidate actions are judged and generated according to signal quality, and the accuracy of phase compensation is improved. The external equipment is controlled in combination with state mapping and trigger execution, the problems of single feature dimension, uncompensated individual difference, uncorrected time delay error, insufficient signal quality control and the like in the existing method are solved, and the stability, accuracy and adaptability of interaction are improved.

The invention provides a 3D Gaussian splash artifact suppression method and system based on high-order regularization and multi-stage pruning, and belongs to the technical field of three-dimensional model reconstruction, and the method comprises the steps: obtaining an initial Gaussian point cloud; and processing the obtained initial Gaussian point cloud by using a pre-trained 3DGS framework to obtain a 3D reconstruction result. According to the method, through polynomial high-order regularization, explicit suppression of high-opacity abnormal points is realized, and the rendering quality is effectively improved. In the rough pruning stage, Gaussian points with low opacity and small contribution are periodically removed to reduce storage, and in the fine pruning stage, spatial abnormal points are further identified and eliminated through joint judgment of local density and anisotropy, so that artifact suppression is realized.

The invention discloses an information data processing system based on brain wave signals, and particularly relates to the field of medical signal processing and brain science application, and the system comprises a data importing and filtering module which supports the input of an original electroencephalogram file and carries out band-pass filtering on the original electroencephalogram file; the bad track detection module is used for automatically detecting and marking bad tracks based on a local outlier factor algorithm, and comparing local density differences between each channel and surrounding channels; the artifact suppression module is used for correcting non-engraving plate transient artifacts in the electroencephalogram; the data restoration and interpolation module is used for carrying out interpolation on the detected bad track and restoring the overall channel layout; the frequency band and ratio characteristic module is used for calculating energy, relative power and ratio of delta, theta, alpha, SMR, beta and betaH frequency bands; the spectrum-space-time characterization module is used for constructing the features into two-dimensional tensors; and the judgment module is used for outputting the mental state category and the attention score. The mental state can be identified based on the juvenile electroencephalogram.

The present application belongs to the field of CT tomographic reconstruction technology and artificial intelligence, and discloses a limited-angle CT reconstruction artifact suppression method based on a multi-domain feature fusion network. In view of the problem that the reconstruction result of traditional CT scanning under limited-angle conditions is prone to artifacts and structural distortion, thereby affecting the image quality and defect detection accuracy, the present application constructs a multi-domain feature fusion artifact suppression network, takes the limited-angle reconstruction result as input, and realizes artifact suppression and detail recovery through the synergistic effect of the encoder part, the decoder part, the feature enhancement part and the feature conversion part. The present application can obtain high-quality tomographic images under limited-angle conditions, effectively reduces the scanning angle and time of industrial CT detection, improves the imaging clarity and reliability without increasing the radiation dose, is suitable for industrial detection of complex structure workpieces, and has important industrial application value.

A computational pupillometry system comprises an imaging device configured to capture video frames of a subject's eye and processors executing instructions to perform advanced pupillary assessment. The system employs multi-frame integration techniques, including super-resolution algorithms that utilize sub-pixel shifts between frames, temporal averaging for noise reduction, and parallax-based artifact mitigation to enhance measurement accuracy. Artificial intelligence models, including temporal neural networks, analyze the enhanced pupillary data to determine pupillary parameters and calculate a light-invariant Pupil Reactivity (PuRe) score. The system processes ambient lighting conditions through computational models that analyze video frames before and after controlled stimulation, enabling consistent scoring across varying environmental conditions. Quality assurance mechanisms provide pre-recording and post-recording validation with real-time feedback. The system integrates with electronic medical records through standardized healthcare protocols and supports synchronized, multi-device deployment across healthcare networks.

The application relates to a mode filter, a topology optimization method and an ultrasonic guided wave damage imaging artifact suppression method, relates to the field of ultrasonic detection and imaging, and the topology optimization method comprises the following steps: acquiring initial design parameters, generating an initial population, each chromosome individual in the initial population is coded by a binary logic matrix representing material distribution and meeting design constraints; a population is optimized by using a genetic algorithm, an adaptability function in the genetic algorithm is constructed based on a mode purity index and an energy transmission index, the mode purity index is the ratio of in-plane displacement integral in a transmission area in finite element simulation with the mode filter to total displacement integral, and the energy transmission index is the ratio of relative energy at a transmission end in the finite element simulation with the mode filter to relative energy of an S0 mode; and final optimized material distribution of the mode filter is output. Compared with the prior art, the application has the advantages of effectively suppressing background artifacts, giving consideration to mode regulation performance and substrate structure integrity and the like.

The present invention relates to vibration correction. A vibration artifact suppression method 100 includes: a) step 110 of reconstructing a three-dimensional image of an object of interest based on projection data acquired by an X-ray imaging system; b) step 120 of determining whether the reconstructed three-dimensional image includes a metal object; and c) step 130 of performing a vibration correction method based on a metal object segmented in the reconstructed three-dimensional image in response to the determination that the reconstructed three-dimensional image includes a metal object; or c2) step 140 of performing a vibration correction method based on an object of interest segmented in the reconstructed three-dimensional image in response to the determination that the reconstructed three-dimensional image does not include a metal object. Using the vibration artifact suppression method, the number of iterations is significantly reduced, resulting in a significant speedup of the overall reconstruction time.

This invention relates to a method for suppressing artifacts in neurally modulated stimuli and reconstructing effective features of a brain-computer interface. The method involves simultaneously acquiring multi-channel neural electrical signals and stimulus event information; establishing a preset time window centered on the event moment; extracting transient evidence features, saturation evidence features, and recovery evidence features that occur synchronously with the stimulus from the neural electrical signals; arbitrating the credibility of the stimulus event based on the event information and evidence features to obtain the event arbitration type; generating a corrected event moment when a time offset is detected to form a credible event table; determining the artifact suppression mode and its application window according to the event arbitration type, whereby the artifact suppression mode includes strong suppression, weak suppression, or bypass processing, and performing artifact suppression or bypass processing on the corresponding neural electrical signals; extracting effective features of the brain-computer interface after suppression; and reconstructing features for the feature gaps caused by suppression according to the event arbitration type, outputting the reconstructed features and credibility markers.

The application discloses a kind of diffusion prior-based adversarial generative remote sensing image super-resolution method, belong to remote sensing image processing technical field.The application divides super-resolution reconstruction into two independent stages of degradation removal and information regeneration: first, through degradation removal network, low-resolution remote sensing image is executed to carry out degradation suppression processing such as denoising, deblurring, artifact suppression, eliminate the content-independent degradation interference caused by imaging and transmission;Again, the degraded removal image is input into the generation network to realize high-resolution information regeneration, the generation network is initialized using the pre-trained diffusion model parameters to have the prior distribution characteristics of remote sensing image, and the content consistency and detail authenticity of the generated image are ensured by combining the adversarial optimization of the discriminant network and the pixel-sensing joint fidelity constraint in the training stage;Inference stage only needs one forward mapping to output high-resolution results, while supporting visual language model semantic text condition guidance, to realize controllable adjustment of generation results.

This invention discloses a method for determining exponential distribution delay based on light intensity and event threshold control, comprising the following steps: Based on the fact that DVS image sensor events exhibit exponential distribution characteristics in the time domain when image information changes rapidly in a scene, an exponential distribution model is established that correlates the time constant with light intensity and a threshold; according to the exponential distribution model and the parameters calculated by the model, the exponential distribution delay of the DVS image sensor event to be analyzed in the time domain is determined, which is characterized by the event trigger rate, including the delay magnitude and its distribution in the spatial domain. This invention, based on the exponential distribution delay model, can quickly and conveniently process and calculate the exponential distribution delay of pixels, providing a basic explanation for the origin of motion artifacts in low-light conditions, and assisting researchers in developing artifact suppression methods based on delay mechanisms.

The present invention relates to a system and method for real-time screening of Attention Deficit Hyperactivity Disorder using electroencephalographic signals processed through a GPU-accelerated time-frequency inference architecture. The invention enables continuous acquisition of multi-channel electroencephalographic data, adaptive preprocessing for artifact suppression and signal stabilization, and parallel execution of time-frequency transformations to extract neurologically relevant features in real time. Extracted features are analyzed using an inference process configured to identify neurological patterns associated with Attention Deficit Hyperactivity Disorder, while continuous validation of signal quality and temporal consistency ensures diagnostic reliability. The system dynamically adapts analytical parameters based on evolving signal characteristics and regulates computational workload to achieve energy-efficient operation during prolonged monitoring. The invention provides a technically integrated, machine-implemented screening solution capable of delivering reliable, real-time neurological assessment suitable for clinical environments.

The invention discloses a PS-OCT polarization artifact suppression method based on a cross coupling coefficient. Comprising two parts of pre-measurement based on matrix calculation and PS-OCT data processing based on a cross coupling coefficient, according to the pre-measurement, a dual-channel interference signal is obtained by scanning a wave plate through a PS-OCT system, and then a ghosting artifact displacement set and the cross coupling coefficient are obtained; in the data processing step, a PS-OCT data processing method based on a cross coupling coefficient is used for detecting a sample and processing a dual-channel interference signal, and finally a PS-OCT image without polarization artifacts is obtained. Through the cross coupling coefficient, the cross coupling effect generated by various polarization elements can be quantified, and accurate suppression of polarization artifacts is achieved; the cross coupling coefficient is obtained in advance through the detection wave plate, in actual imaging, the cross coupling coefficient does not need to be measured any more, and the speed of suppressing polarization artifacts is increased; on the premise that the original configuration of the system is not changed, the polarization artifacts are comprehensively suppressed.

The invention relates to a mode filter, a topological optimization method and an ultrasonic guided wave damage imaging artifact suppression method, and relates to the field of ultrasonic detection and imagines.The topological optimization method comprises the steps that initial design parameters are obtained, and an initial population is generated; each chromosome individual in the initial population is formed by coding a binary logic matrix which represents material distribution and meets design constraints; a genetic algorithm is adopted for population optimization, a fitness function in the genetic algorithm is constructed based on a mode purity index and an energy transmission index, and the mode purity index is a ratio of a transmission region in-plane displacement integral to a total displacement integral under finite element simulation added with a mode filter; the energy transmission index is a ratio of relative energy of a transmission end under finite element simulation added with a mode filter to relative energy of an S0 mode; and outputting the final optimized material distribution of the mode filter. Compared with the prior art, the method has the advantages that background artifacts are effectively inhibited, and the mode regulation and control performance and the matrix structure integrity are both considered.

A physiological mechanical signal monitoring system and method utilizing motion state to suppress artifacts, belonging to the field of biomedical signal monitoring and processing technology, aims to improve the existing distributed multi-node vibration and physiological signal acquisition systems, which suffer from severe crosstalk between multiple channels, signal aliasing caused by artifacts, and feature distortion, failing to meet the requirements of high-fidelity, multimodal dynamic acquisition. Key technical points: The signal monitoring system includes a multi-point distributed acquisition module, a motion state recognition module, an artifact detection and adaptive filtering module, and a main control and signal fusion module. The main control and signal fusion module is used to fuse and reconstruct the multi-channel physiological mechanical signals after artifact suppression processing, outputting high-fidelity physiological signals. This invention significantly reduces the coupling effect between different channels caused by environmental vibration, body motion impact, and matrix resonance through independent support of multiple spatial nodes and isolation of mechanical paths, significantly improving the spatial discrimination ability and original independence of the signals. This invention achieves multi-point physical isolation at the structural level, and then combines vector superposition to achieve preliminary noise reduction, effectively reducing artifacts caused by non-target actions and environmental disturbances.

This invention provides a method, system, device, and storage medium for extracting line art from anime images, belonging to the field of image processing technology. The method includes: first, performing preprocessing operations on the input artistic image; then deploying a multi-branch feature extraction network, utilizing parallel convolutional paths and inter-layer skip connections to collaboratively capture multi-scale edge response features, constructing a hierarchical feature map with spatial awareness. Next, a dynamic feature fusion module is designed to optimize cross-layer features in both local receptive field feature aggregation and pixel-level semantic alignment, completing the fusion process with learnable attention weights. Finally, a joint supervision mechanism is established, fusing an improved cross-entropy loss function and background artifact suppression constraints, improving line generation quality and suppressing background noise through pixel-level supervision. This method can achieve end-to-end conversion from complex artistic images to structured line art, effectively eliminating artifact interference while maintaining line continuity, and outputting editable line art that meets professional creative needs.

The invention provides a CBCT image tooth segmentation method based on deep learning, and relates to the technical field of oral cavity image segmentation, and the method comprises the steps: extracting a two-dimensional slice from a CBCT image as an input image; carrying out normalization, artifact suppression and data enhancement on the image to improve the preprocessing quality; constructing a segmentation network structure fusing self-calibration multi-scale attention and a mixed attention mechanism, wherein the segmentation network structure is used for extracting features and enhancing context information; training a model based on the tooth image data set, and performing parameter learning by adopting a joint loss function and an optimizer; and finally, tooth segmentation is realized through the trained model, and three-dimensional reconstruction is carried out. The method has the advantages of high segmentation precision, high processing speed, light structure and the like, and is suitable for automatic identification and diagnosis assistance of teeth in stomatology.

The invention belongs to the field of CT (Computed Tomography) tomography reconstruction technology and artificial intelligence, and discloses a finite angle CT reconstruction artifact suppression method based on a multi-domain feature fusion network. In order to solve the problem that image quality and defect detection precision are affected due to the fact that a reconstruction result of traditional CT scanning is prone to artifacts and structural distortion under the condition of a limited angle, a multi-domain feature fusion artifact suppression network is constructed, and a limited angle reconstruction result is used as input. Through the synergistic effect of the encoder part, the decoder part, the feature enhancement part and the feature conversion part, artifact suppression and detail recovery are realized. According to the method, high-quality cross-sectional images can be obtained under the limited angle condition, the scanning angle and time of industrial CT detection are effectively reduced, the imaging definition and reliability are improved on the premise that the radiation dosage is not increased, and the method is suitable for industrial detection of workpieces of complex structures and has important industrial application value.

The application discloses a multi-modal medical image intelligent fusion diagnosis process, which comprises the following core process steps: S1, multi-modal image data acquisition and preprocessing, collecting original data of at least two kinds of clinically commonly used image modalities, performing format standardization, artifact suppression, gray scale normalization and quality screening processing on the original data, and obtaining standardized image data; S2, cross-modal adaptive registration and fusion integrated processing, based on a cross-modal registration and fusion integrated network, synchronously realizing spatial alignment and preliminary feature fusion of the standardized image data, and generating an aligned feature map; and S3, layered feature extraction, adopting a differentiated feature extraction architecture adapted to different modal imaging characteristics.The application realizes the transparency and traceability of the diagnosis process through feature contribution degree visualization and diagnosis logic tracing, enables a clinician to clearly understand the diagnosis basis, meets the requirement of clinical diagnosis and treatment on explainability, significantly improves the trust degree of the clinician on the diagnosis result, and reduces the clinical application threshold.

The invention discloses an HDR image reconstruction method based on spatial convolutional coding and dynamic convolution. The method comprises the steps of obtaining an LDR image data set, constructing an HDR image reconstruction network, training the HDR image reconstruction network and testing the HDR image reconstruction network. Due to the fact that the space convolution coding module, the dynamic convolution space attention module and the linear cross frame UFO module are adopted, the problems that in the prior art, artifacts are caused by large motion of an object in a dynamic scene, and the detail and color recovery precision is insufficient under the complex illumination condition are solved. Experimental results show that the method significantly improves the HDR reconstruction quality in a dynamic scene, and has strong reconstruction capability in the aspects of color fidelity and visual perception quality. The method has the advantages of high artifact suppression capability, high detail and color recovery precision, high adaptability and the like, and is suitable for the technical fields of mobile phone photographing, photography, screen display, video monitoring, virtual reality and the like.

The invention discloses a magnetic resonance training data generation method and system based on GPU acceleration Bloch simulation and combined with modal transformation, and relates to medical imaging. The method comprises the following steps: establishing a multi-parameter virtual object library, and defining voxel-level T1, T2, PD and B0 / B1 distribution and coil sensitivity; according to an input sequence and acquisition parameters, time domain simulation is carried out by adopting GPU parallel Bloch solution, and non-ideal factors are injected as required; reconstructing to obtain k space and image domain data and generating paired labels; target comparison, organs and scenes are expanded through rule-driven or data-driven modal transformation; samples and metadata are packed to support traceability and verification. The system is composed of a parameter configuration module, a virtual object library module, a GPU simulation module, a reconstruction module, a modal transformation module and a data set management module. On the premise that physical consistency is guaranteed, data generation efficiency and diversity are remarkably improved, and the method is suitable for downstream tasks such as reconstruction, artifact suppression and quantitative imaging.