164 results about "Motion field" patented technology

A system for dynamic latent space adaptation using spatiotemporal kernel context for multiscale rendering with hierarchical and Lorentzian autoencoders. The Spatiotemporal Kernel Estimator (SKE) analyzes media through motion field, temporal recurrence, frequency band, and scene semantics analyzers to generate adaptive kernel parameters encoding content-specific importance distributions. The system dynamically adapts latent manifold geometry by modifying metric tensor properties according to kernel context, enabling content-aware compression that allocates representational capacity based on visual significance. A multiscale cache implements kernel-adaptive retention policies prioritizing important regions. An adaptive renderer provides intelligent level-of-detail selection based on zoom level and kernel-estimated importance, optimizing processing allocation. The self-optimizing architecture continuously refines kernel context and geometric adaptation based on user interaction and performance feedback, achieving superior compression ratios and perceptual quality. Applications include bandwidth-efficient video streaming, virtual reality, scientific visualization, and cognitive video analytics requiring intelligent context-aware visual processing.

The invention relates to the technical field of computer vision, in particular to an intelligent conference video frame dynamic coding method based on multi-modal semantic understanding, which comprises the following steps: acquiring a video stream sequence and a synchronous audio stream in a conference scene in real time; performing semantic analysis and decoupling on the video stream sequence, and extracting key frames and subsequent frames; extracting a sparse motion field from a subsequent frame, and segmenting a video frame into candidate visual areas including a face, a mouth shape and a background; extracting audio semantic features, executing cross-modal semantic correlation analysis, calculating semantic correlation between the sparse motion field distribution features and the audio semantic features, and positioning a pronunciation area highly related to the voice content; and calculating a quantization offset value of each candidate visual area according to the semantic relevancy, applying the quantization offset values in different areas, and packaging the quantization offset values into a variable-code-rate video code stream. According to the invention, the multi-mode semantic understanding model is constructed to carry out deep semantic analysis on the video frame content so as to realize the dynamic coding of the conference video frame.

The invention relates to the technical field of face verification, and discloses a face identity verification data processing method based on dynamic feature extraction, which comprises the following steps: acquiring a continuous face video frame sequence and calculating a full-pixel instantaneous velocity vector to generate an original dense optical flow field, selecting rigid region anchor points to calculate a rigid affine transformation matrix and construct a theoretical rigid motion field, performing differential stripping on the theoretical rigid motion field from the original dense optical flow field, and extracting a non-rigid micro-motion residual field; mapping the non-rigid micro-motion residual field to a facial muscle topological grid to generate a time sequence feature tensor, and calculating a geodesic line distance between a covariance matrix of the time sequence feature tensor and a reference dynamic feature in a Riemannian manifold space; when the geodesic distance is smaller than a threshold value, verification is passed, through a rigid-non-rigid orthogonal decomposition mechanism, the special viscoelastic micro-motion and cooperation law of biological soft tissue is captured by utilizing a residual field, and the high-simulation mask is effectively defended.

The invention discloses an event-image dual-mode fusion video turbulence correction method, medium and system, and belongs to the field of digital image processing, and the method comprises the steps: synchronously obtaining an event voxel of a target and a to-be-corrected image sequence; inputting the event voxels into a pre-selected coding and decoding structure to obtain features, projecting the features along the optical flow direction, and performing three-dimensional reconstruction to obtain target motion features; extracting background scene representation; fusing the target motion feature and the background scene representation in a channel dimension to obtain an edge guide feature; inputting the image sequence and the image sequence into a pre-trained video restoration network to obtain a turbulence-corrected video; the training loss of the coding and decoding structure comprises an optical flow field estimated according to an image sequence without turbulence disturbance and a target motion field obtained by projecting the feature along the optical flow direction. The method can accelerate the recovery process and improve the recovery quality.

The invention discloses a multi-modal deep reinforcement learning gas leakage decision optimization method based on optical flow, and the method comprises the steps: carrying out the constraint and correction based on a pixel motion field in combination with wind speed estimation, and obtaining an optical flow reconstruction wind field; based on plume probability distribution, combining with an image intensity and radiation transmission model, obtaining concentration field estimation, fusing gas concentration data and an optical flow reconstruction wind field, and forming full-field state description; establishing a convective diffusion model, predicting the concentration field at the next moment through the convective diffusion model based on the full-field state, establishing a convective diffusion model, and predicting the concentration field at the next moment through the convective diffusion model based on the full-field state. According to the technical scheme of the invention, the dynamic evolution precision of gas leakage can be effectively improved by introducing the wind field reconstruction method of the optical flow and the scatter-free constraint, the prediction stability is improved, the physical consistency of the wind field and the concentration distribution is ensured, and stronger data support is provided for emergency control.

The invention provides an endoscope image stabilization control method, which comprises the following steps of: acquiring an original dithering image sequence and inertial data of an endoscope, fusing the original dithering image sequence and the inertial data, acquiring an initial mixed motion field, performing operation semantic segmentation on the original dithering image sequence, and acquiring a collaborative attention map in combination with a physical motion mechanism. Obtaining a space-time adaptive filtering kernel through the collaborative attention map, filtering the initial mixed motion field according to the space-time adaptive filtering kernel, obtaining a stable motion field, carrying out image deformation and interpolation synthesis processing on the stable motion field, and obtaining a stable image sequence; through the integrated technical scheme of operation semantic segmentation, dynamic image stabilization optimization and accurate jitter suppression, the problem of interference of endoscope jitter on the neurosurgery operation visual field is solved.

The invention provides an eye movement tracking method, a control unit and eye movement tracking equipment, and relates to the technical field of eye movement tracking. The method comprises the steps of obtaining a surface image of a target eyeball; extracting composite phase distribution from the surface image, wherein the composite phase distribution represents phase coupling information after the incident light is jointly modulated by a cornea interface and a crystalline lens interface of the target eyeball; based on eyeball geometric constraint and cornea interface smoothness prior, according to the composite phase distribution, constructing a decoupling objective function with separation of cornea phase distribution and crystalline lens phase distribution as a target; taking a minimum decoupling objective function as an optimization objective, and iteratively determining cornea phase distribution decoupled from crystalline lens phase distribution; and determining the visual axis direction of the target eyeball according to the cornea phase distribution. Through the double-interface phase decoupling technology, the visual axis error in a large view field and eyeball depth motion scene is effectively reduced, and the eye movement tracking precision is remarkably improved.

The invention discloses a comprehensive performance testing system and method for a power-assisted exoskeleton, and belongs to the technical field of robot testing. Comprising a multi-modal data acquisition layer, a scenarized test module and an intelligent processing and evaluation layer, the multi-modal data acquisition layer comprises a multi-modal data acquisition module and a hierarchical data synchronization module, and the scenarized test module selects modules in the multi-modal data acquisition module to perform data acquisition under the conditions of wearing an exoskeleton and not wearing the exoskeleton in a corresponding scene according to a pavement scene and a motion scene which need to be tested; and the intelligent processing and evaluation layer outputs a final comprehensive performance index based on a three-layer fusion strategy according to the output of the scene test module. Through multi-source data fusion analysis and multi-dimensional evaluation indexes, the auxiliary effect, the energy consumption efficiency and the use safety of the exoskeleton can be comprehensively and objectively reflected, and accurate guidance is provided for research and development optimization of the exoskeleton.

The motion estimation of an anatomical structure may be performed using a machine-learned (ML) model trained based on medical training images of the anatomical structure and corresponding segmentation masks for the anatomical structure. During the training of the ML model, the model may be used to predict a motion field that may indicate a change between a first training image and a second training image, and to transform the first training image and a corresponding first segmentation mask based on the motion field. The parameters of the ML model may then be adjusted to maintain a correspondence between the transformed first training image and the second training image and between the transformed first segmentation mask or a second segmentation mask associated with the second training image. The correspondence may be assessed based on at least a boundary region shared by the anatomical structure and one or more other anatomical structures.

The invention provides a dysphagia screening system based on computer vision, and relates to the technical field of image recognition, and the system comprises an image collection module which collects image data of a swallowing event to obtain a time sequence image sequence; the motion field establishing module is used for extracting a motion point track sequence according to the time sequence image sequence and establishing a global motion transformation field; the motion separation module is used for carrying out registration correction on the global motion transformation field and carrying out motion separation on the stabilized image sequence to obtain local deformation components; the region segmentation module is used for carrying out key region segmentation on the local deformation components to obtain swallowing function parameters; and the risk identification module is used for identifying the obstacle risk of the swallowing function parameters and generating a swallowing obstacle screening report. According to the method and the device, the technical problem of relatively poor screening accuracy of the dysphagia based on the computer vision in the prior art can be solved, and the technical effect of improving the screening accuracy of the dysphagia based on the computer vision is achieved.

The invention relates to the technical field of computer vision and precise opto-electro-mechanical control, in particular to a motion scene intelligent recognition focus tracking system based on deep learning, which comprises a high-frequency vision acquisition discretization unit, a video stream decoupled into an RGB semantic stream and a gray-scale high-frequency stream, and a high-frequency vision acquisition discretization unit, the RGB semantic stream and the gray high-frequency stream are sent to a double-stream uncertainty prediction unit; the double-current uncertainty prediction unit is used for obtaining a target state vector and generating a transient prediction value; the motion entropy arbitration fusion unit is used for calculating the motion entropy based on the inertial predicted value, the target state vector and the covariance matrix, weighting the inertial predicted value and the transient predicted value based on the motion entropy, and generating a target point and a future reference trajectory; the entropy sensing self-adaptive control unit is used for solving the control voltage to drive the focusing module; according to the invention, focusing lock loss caused by overshoot of the control target is effectively prevented, and the tracking precision under extreme working conditions is ensured.

Projected motion field hole filling includes determining a motion field of a current frame using a block true subset of the current frame, the motion field including, for each block in the block true subset, a respective motion vector projected onto a reference frame. For a block in the current frame that does not have a motion vector in the motion field, a respective motion vector of a spatial domain neighboring block of the proper subset of the block is reused as a projected motion vector of the block within the motion field. A list of motion vector candidates may be determined using the motion field. A reference motion vector for the current block may be selected from the motion vector candidate list. The current block may be encoded into an encoded bitstream or decoded from an encoded bitstream using the reference motion vector.

A system for dynamic latent space adaptation using spatiotemporal kernel context for multiscale rendering with hierarchical and Lorentzian autoencoders. The Spatiotemporal Kernel Estimator (SKE) analyzes media through motion field, temporal recurrence, frequency band, and scene semantics analyzers to generate adaptive kernel parameters encoding content-specific importance distributions. The system dynamically adapts latent manifold geometry by modifying metric tensor properties according to kernel context, enabling content-aware compression that allocates representational capacity based on visual significance. A multiscale cache implements kernel-adaptive retention policies prioritizing important regions. An adaptive renderer provides intelligent level-of-detail selection based on zoom level and kernel-estimated importance, optimizing processing allocation. The self-optimizing architecture continuously refines kernel context and geometric adaptation based on user interaction and performance feedback, achieving superior compression ratios and perceptual quality. Applications include bandwidth-efficient video streaming, virtual reality, scientific visualization, and cognitive video analytics requiring intelligent context-aware visual processing.

The utility model discloses a testing device for simulating a rotary motion imaging scene, and the device comprises a roller assembly which is externally provided with a fixing plate for fixing an object to be imaged; the transmission shaft is fixed at a rotating shaft of the roller assembly; and the driving assembly is used for driving the transmission shaft to rotate so as to drive the roller to rotate. The device can continuously operate when a motion scene is simulated in an imaging test, and the requirement of the imaging test under a long-time continuous stable condition is met.

A sparse helical CT image reconstruction method based on a differentiable helical reconstruction operator. First, actual helical scanning geometric parameters of a subject and corresponding full-angle helical projection data are acquired, and a final reconstructed image is acquired by means of seven steps. In the present invention, actual scanning geometry is used to perform forward projection on a reconstructed sparse-angle image, thereby providing geometric prior guidance for missing projections; moreover, on the basis of the similarity and redundancy characteristics of adjacent projections in helical scanning, a projection completion network is constructed, and by learning bidirectional motion fields of adjacent angles and in combination with geometric prior projections, intermediate missing projection data is jointly synthesized; in addition, the global streak artifact restoration of the image is realized; and finally, the joint training of a projection domain and an image domain is realized, thereby facilitating integral restoration by using projection-image dual-domain information, and data collected within two pitches is used for restoration, thereby effectively avoiding an excessive computational load.

The embodiment of the invention provides a depth video frame interpolation method and device based on occlusion perception and dynamic optimization, a storage medium and electronic equipment, and relates to the field of video frame interpolation, and the method comprises the steps: obtaining two frames of target images to be interpolated, a pre-calculated occlusion image and an optional reference intermediate frame, carrying out normalization and edge filling preprocessing on the target image and the shielding image; based on the preprocessed target image and the potential spatial dimension of the diffusion model, initializing random noise capable of being subjected to gradient optimization as a potential mark; designing an iterative optimization framework, and performing dynamic optimization on the potential mark by using the preprocessed target image and the shielding image; and inputting the optimized potential mark into a diffusion model to generate a final intermediate frame, and restoring the intermediate frame to the size of the original target image through reverse filling operation to complete frame insertion. According to the method, the core problem that blurring and artifacts are easy to generate when shielding areas and complex motion scenes are processed in the prior art is solved.

The application provides a ball game project general training method and an auxiliary referee system based on a deep learning technology for motion target detection, and comprises the following steps: establishing a basic weight parameter, establishing a trained weight parameter through deep learning, optimizing a parameter of a specific motion scene, collecting actual motion scene images through a designed parameter optimization module when an error occurs in the basic weight parameter, performing optimization training, and establishing an optimized weight parameter of the specific scene. The application establishes a gridized motion index parameter, feeds back a grid motion index change in real time during training or competition, and assists a coach in optimizing or improving a training effect according to the grid motion index change. An integrated technology of a multi-camera which can be arbitrarily stacked and expanded can be used for auxiliary refereeing during competition, and can also be used for playing back high-speed high-definition images of any viewing angle during training to help athletes analyze action postures and the like.

The invention relates to the field of intelligent medical treatment, in particular to a continuous time sequence CT image construction method and device based on EIT and a program product. Comprising the steps of performing CT scanning on a to-be-detected part and a to-be-detected person, and obtaining CT images of the to-be-detected person at the end of inspiration and the end of expiration and EIT impedance images of the same continuous time period from the end of inspiration to the end of expiration; the CT image at the end of inspiration and the CT image at the end of respiration are used as a first frame and a tail frame respectively, and a motion field for converting the CT image at the end of inspiration into the CT image at the end of expiration is predicted based on the dynamic change of the EIT impedance image in a continuous time period from the end of inspiration to the end of expiration; registering the motion field, the first frame and the tail frame to obtain a registered motion field, a registered first frame and a registered tail frame; and generating a CT image in a middle time period through the registered motion field, the first frame and the tail frame, and obtaining a CT image in a continuous time period from the end of inspiration to the end of expiration. The application has good clinical value.

The invention relates to the technical field of computer vision and video image processing, and discloses a video picture intelligent frame supplementing method and system based on deep learning, and the method comprises the steps: carrying out the feature coding and semantic segmentation of front and rear frames of a video sequence, and dividing an image into a rigid body instance region and a non-rigid body region; constructing a parameterized rigid body reference flow field by predicting geometric transformation parameters of the rigid body instance; calculating a space gradient and a semantic edge weight of the reference flow field; a motion field gradient consistency constraint is introduced when a residual optical flow is predicted, and a semantic edge weight is used for forcibly synthesizing a space gradient of an optical flow field to keep consistent with a reference flow field at the edge; and determining a pseudo depth sequence according to the semantic category, and executing hierarchical forward transformation and hole filling by using the synthetic optical flow to generate a target intermediate frame. According to the method, parameterized modeling and a gradient constraint mechanism are combined, the problems of structural deformation and obscured shielding edges in a rigid body large-displacement scene are solved, and the definition and structural integrity of a complementary frame picture are improved.

The invention relates to the technical field of video data, and discloses a compression extraction and processing method for key frames in video data, which comprises the steps of input and preprocessing, inter-frame difference score calculation, optical flow motion score calculation, fusion scoring, time sequence smoothing and local extremum screening, redundancy suppression, and key frame compression and index reconstruction. Through the steps of inter-frame difference analysis, optical flow motion field calculation, adaptive fusion scoring, time sequence smooth constraint, redundancy suppression and feature compression encoding, adaptive extraction and compression processing are performed on frames with significant changes or semantic importance in a video sequence, and low-layer change features are obtained through pixel layer frame difference calculation. Capturing local and global motion features through optical flow vector field modeling; constructing a criticality scoring model, carrying out weighted fusion on multi-channel features, carrying out smoothing and extreme value detection on a time dimension, and selecting a most representative video key frame; compression and index structure reconstruction are completed, and structured storage and efficient restoration of video data are achieved.

The invention discloses a motion scene human body posture estimation method driven by adaptive time modeling and a medium, and the method comprises the steps: constructing an adaptive time modeling motion scene 3D human body posture estimation network model based on the motion speed degree; the robustness of the model to a motion scene is improved by using motion prior based on physical constraints and predicting the uncertainty of 3D human body articulation points; human body motion videos with different motion speeds are used as input, a corresponding 2D human body joint point sequence is obtained on the basis of an existing high-quality 2D human body posture estimation model, the 2D human body joint point sequence is input into the network model, and a trained model is obtained; according to the method, the receptive field can be adaptively adjusted according to the movement speed, the small receptive field is used for capturing transient changes for fast movement, the large receptive field is used for capturing long-term dependence for slow movement, the method adapts to movement at different speeds, and therefore the human body posture estimation accuracy in movement scenes at different speeds is improved.

The application discloses a target tracking method, device and storage medium. The target tracking method comprises the following steps: acquiring position information of a target object, querying scene modeling data matched with the position information in a preset scene modeling database, and acquiring real-time images collected by an image collection device matched with the position information, so as to accurately call resources according to the position of the target object, improve subsequent modeling efficiency, then, the scene modeling data and the real-time images are fused to obtain more accurate and convenient observation real scene modeling results, and the target object is tracked based on the real scene modeling results, so that more image details and visual ranges are provided for security, tracking and other scenes on the premise that the target object is not lost, and comprehensive research and judgment on the motion scene of the target object are facilitated.

The invention relates to a method for determining an exposure time (t) of a camera (1) of a camera system (10) for preferably moving and / or for machine object recognition in a moving scene, in which a predefined minimum angular image resolution (Ra *) is followed, comprising the following steps: a) determining an effective angular image resolution (Ra, eff (t)) as a function of the exposure time (t), b) the maximum allowable exposure time (tmax) is determined as the exposure time when the effective angle image resolution (Ra, eff (t)) is the value of the minimum angle image resolution (Ra *), and c) the exposure time (t) is set to a value less than or equal to the maximum allowable exposure time (tmax). The invention also relates to a camera system (10).

This invention discloses a video turbulence correction method, medium, and system based on event-image dual-mode fusion, belonging to the field of digital image processing. The method includes: simultaneously acquiring event voxels of the target and the image sequence to be corrected; inputting the event voxels into a pre-trained encoding / decoding structure to obtain features; projecting the features along the optical flow direction and performing 3D reconstruction to obtain target motion features; extracting background scene representations; fusing the target motion features and background scene representations in the channel dimension to obtain edge-guided features; and inputting the image sequence and background scene representations together into a pre-trained video restoration network to obtain a turbulence-corrected video. The training loss of the encoding / decoding structure includes: representing the optical flow field estimated from the image sequence without turbulence disturbance, and representing the target motion field obtained by projecting the features along the optical flow direction. This invention can accelerate the restoration process and improve restoration quality.

The invention discloses a multi-modal data analysis method in a motion scene, electronic equipment and a storage medium, and belongs to the technical field of data processing. The method comprises the following steps: acquiring first data in a preset time period; obtaining a first feature vector sequence and a second feature vector sequence; fusing the first feature vector sequence to obtain a first fusion vector; fusing the second feature vector sequence to obtain a second fusion vector; obtaining first weight distribution corresponding to the first fusion vector and second weight distribution corresponding to the second fusion vector based on an anomaly discovery algorithm; and determining a classification label of the state according to the first weight distribution and the second weight distribution. The technical problem that the accuracy of a monitoring result is low due to the fact that a sensor signal is interfered in a bumping or moving scene can be solved. The method is mainly used for body state monitoring. According to the invention, cross validation is carried out through the first weight distribution and the second weight distribution, and the accuracy is improved.

Systems and techniques for processing video data are provided. For example, a device can obtain one or more first sets of collocated motion vector data and one or more second sets of collocated motion vector data associated with respective first and second blocks of video data included in a current frame of video data. The device can project the one or more first sets of collocated motion vector data into a first projected motion field associated with a first buffer and project the one or more second sets of collocated motion vector data into the first projected motion field associated with the first buffer. Based on the one or more first and second sets of projected collocated motion vector data, the device can decode the first block of video data based on the first projected motion field associated with the first buffer.

A virtual eye test for evaluating dynamic visual acuity can be conducted in a virtual reality (VR) environment. The test utilizes an electronic device equipped with a head-mounted display (HMD) and a camera. The device generates a VR user interface corresponding to a three-dimensional virtual environment and renders it on the HMD. Real-world motion and target recognition visual tasks are simulated within the VR interface. Using the camera, the device tracks eye movements and response times to visual stimuli presented during these tasks. Dynamic visual acuity is measured based on the tracked eye movements and response times, providing a comprehensive assessment of visual performance in motion-based scenarios.

The application discloses a kind of visual detection methods based on YOLO multimodal space-time alignment, belong to industrial visual detection technical field.The method is by hardware trigger synchronous acquisition first modality image and second modality image, and respectively constructs feature block sequence;The state transition of second modality is dynamically gated modulation using the hidden state of first modality, realize cross-modal space-time alignment and feature fusion;Then, the fusion feature is input into bidirectional visual Mamba backbone network to extract multi-scale global features, and the YOLO detection head is combined to complete target positioning and class recognition.The application can effectively improve the accuracy, stability and real-time performance of multimodal visual detection in high-speed motion scene.

This invention relates to the field of social network user positioning technology, and particularly to a method and system for locating the start and end points of user motion trajectories based on probe sensing range constraints and coordinate mapping. The method involves determining platform probe sensing range constraint parameters, deploying probes in a target area based on these parameters to obtain a user trajectory map of the target area. This user trajectory map is overlaid as a data layer on a background map with a road network structure as a reference. Pixel coordinates of the user's start and end points and road intersections are extracted from the user trajectory map. Based on the mapping relationship between road intersection pixel coordinates and geographic latitude and longitude, a conversion function from the current user trajectory map pixel coordinates to geographic latitude and longitude is determined, and the geographic latitude and longitude of the user's start and end points are obtained using this conversion function. This invention enables high-precision and high-efficiency positioning of user motion start and end points on social motion platforms that use motion trajectory images to carry user location information. It exhibits strong robustness and good generalization ability, and can provide technical support for motion scene service optimization and network security supervision.

Devices, methods, and non-transitory program storage devices are disclosed herein to perform intelligent determinations of non-linear (i.e., dynamic) image recording rates for the production of improved timelapse videos. The techniques described herein may be especially applicable to timelapse videos captured over long durations of time and / or with varying amounts of device motion / scene detail change over the course of the captured video (e.g., when a user is walking, exercising, driving, etc. during the video's capture). By smoothly varying the image recording rate of the timelapse video in accordance with estimates of device motion, the quality of the produced timelapse video may be improved (e.g., fewer long stretches of the video with no action or too little action, as well as fewer stretches of the video where there is so much rapid action in the timelapse video that it is difficult for a viewer to perceive what is happening in the video).