426 results about "Motion estimate" patented technology

A method for simultaneous localization and mapping underwater. A robot is equipped with an IMU inertial unit and a sonar unit. When the robot submerges underwater, a buoy connected to the robot floats on the water surface, and the buoy moves in coordination with the movement of the robot. The position and observed velocity of the buoy are obtained by a shore-based lidar. During motion estimation in a SLAM algorithm, when an angular velocity is below a preset angular velocity threshold, the observed velocity of the buoy is decomposed into x-axis velocity and y-axis velocity, and updated as the two-dimensional operating velocity of the robot. The polar coordinates of an obstacle in a map under the coordinate system of the robot are associated with the polar coordinates of sonar data of a current frame by using the Mahalanobis distance. The polar coordinates of a new obstacle are converted to the world coordinate system and added to the map. The present invention has the advantages of not relying on underwater visibility, not requiring pre-installed devices, and possessing good generalization and stability.

Methods and systems implement a pleno-generation face video compression framework with bandwidth intelligence for generative models and compression. Heterogeneous-granularity facial description regularizes long-term dependencies between video frames and compensates for motion estimation errors caused by compact representations of motion information. A generative decoder reconstructs heterogeneous-granularity visual representations, providing auxiliary visual signals for attention-based recalibration of a GFVC-reconstructed face signal. A coarse-to-fine generation strategy avoids error accumulation. High efficiency for heterogeneous-granularity signal compression is achieved by two different entropy-based signal compression methods: heterogeneous-granularities feature representation from the key-reference frame as hyperpriors to optimize the entropy model for compressing heterogeneous-granularity feature from subsequent inter frames, and a feature difference operation for heterogeneous-granularities feature representation between key-reference and subsequent inter frames, such that the entropy model only compresses heterogeneous-granularities feature residual for redundancy reduction. Mixed-model dataset generation and training and model-specific dataset generation and training are also provided.

The invention relates to the technical field of map construction, and provides a mapping method and device based on fusion of a laser radar and an inertial measurement unit. According to the scheme, motion distortion compensation is carried out on original point cloud data through motion estimation based on inertial data, the compensated point cloud is matched with a local map, and the local map is obtained through a filter; carrying out tight coupling fusion on the inertial data and the point cloud matching result, and outputting the pose information of the sensor at the current moment; according to the pose information of the sensor, the compensated point cloud is added into the dynamically managed local map, and the updated local map is periodically published, so that a complete mapping function from data acquisition, processing and fusion to map construction is realized. And through motion distortion compensation, the inertial data and the point cloud matching result are tightly coupled and fused, and the updated local map is periodically published, so that high mapping precision can still be kept when the mapping area is large.

The invention belongs to the technical field of multi-camera cooperative imaging, and discloses a multi-camera cooperative anti-shake control system and method. The method comprises the following steps: generating a global hardware trigger signal by using an FPGA (Field Programmable Gate Array), synchronizing IMU and multi-camera acquisition, and combining motion parameters to obtain a space-time aligned multi-view image; obtaining a global jitter component through feature extraction, motion vector field generation, depth map analysis and the like; fusing the data with IMU data to generate a trajectory, and inputting the trajectory into LSTM to obtain future compensation amount; and cutting based on deep layered transformation to obtain an EIS optimized frame, dynamically electing a main camera for super-resolution reconstruction, and outputting a final anti-shake video stream. Through multi-source information fusion, layered precise compensation and dynamic optimization, the accuracy and stability of motion estimation are remarkably improved, the jitter influence of the finally output video stream is effectively weakened, and the reliability and visual coherence of the image stabilization effect in a complex scene are ensured.

The invention provides a medical image denoising and segmentation integrated model trained based on a conductible diffusion method, and belongs to the field of medical image processing, and the model comprises an improved denoising and diffusion module which is used for receiving an initial medical image, predicting an original clean signal of the initial medical image based on an improved denoising and diffusion probability model, and obtaining a final denoised medical image; the joint motion and segmentation module is used for receiving the initial medical image, extracting features through a shared encoder, and performing joint learning through a motion estimation branch and a segmentation branch to obtain segmentation data; the cascade training mechanism carries out end-to-end training on the improved de-noising diffusion module and the joint motion and segmentation module through a derivable connection, gradient back propagation is realized by using a joint loss function, and cascade training is completed. According to the method, the problem that the quality of segmented medical images is reduced due to the fact that denoising and segmentation tasks cannot be collaboratively optimized due to the fact that a traditional denoising method cannot be guided in sampling and is difficult to carry out cascade training with a segmentation model is solved.

The invention provides an imaging spectrometer motion compensation imaging model construction method, an imaging model and a device, and relates to the technical field of space remote sensing. The method comprises the following steps: establishing a satellite orbit model, and defining a plurality of coordinate systems such as a detector coordinate system and a spectrometer imaging coordinate system and a conversion relation thereof; constructing a detector-scanning mirror-lunar surface ideal geometric mapping model based on the load optical parameters and a ray tracing method; determining the relation between the rotation angle of the scanning mirror and imaging spectrum view field positioning by using a reflecting prism conjugation theory; a quantitative relation is established by combining parameters such as satellite orbits and position postures, and finally a motion compensation imaging model is established. According to the method, the influence of the on-orbit error on the compensation imaging model is analyzed, and the error is used as an uncertainty component to be coupled into the model. According to the method, the problems of low model precision, neglect of influence of orbit observation parameters on imaging and the like in the prior art are solved, high-precision motion estimation and compensation can be realized, and the quality of data acquired by an imaging spectrometer is remarkably improved.

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.

A computer-implemented method includes converting by a pose tokenizer, based on a learned codebook, pose parameters of a body into a sequence of discrete pose tokens; randomly masking a portion of the sequence of discrete pose tokens; predicting the randomly masked sequence of discrete pose tokens based on multi-scale features extracted from a monocular image by an image conditioned masked transformer; optimizing the sequence of discrete pose tokens by aligning a re-projected three-dimensional (3D) pose with an estimated two-dimensional (2D) pose; directly regressing, from the multi-scale features, a shape parameter of the body and a weak perspective camera parameter; and generating a 3D mesh reconstruction of the body based on the shape parameter and the weak perspective camera parameter.

The invention discloses a motion carrier self-motion estimation method based on multi-radar fusion, a carrier and a medium, and relates to the technical field of motion carrier positioning. The method comprises the following steps: performing consistency check among radars based on self-motion estimation results of the radars, and judging whether the radars with abnormal self-motion estimation results exist or not according to check results; if not, the self-motion estimation results of all radars are fused, and a final self-motion estimation result is output; if so, identifying an abnormal radar from the radars according to at least one quality index; and discharging the self-motion estimation results of the radars identified as abnormal, and based on the self-motion estimation results of the remaining radars not identified as abnormal, performing fusion or directly outputting the self-motion estimation results as the final self-motion estimation result. According to the method, the abnormal source can be automatically identified and isolated, and the robustness and the function security level of the system in a real complex environment are remarkably improved.

The invention discloses an unmanned aerial vehicle detection and tracking method based on a multi-modal image, and belongs to the field of target detection and tracking. The unmanned aerial vehicle in the air is detected through the combined action of multiple modes, so that the reliability of unmanned aerial vehicle detection in various environments is realized, and identification and classification of the unmanned aerial vehicle are realized; in the tracking stage, a prediction structure in which LSTM and a Kalman filter are fused is introduced, so that the state modeling capability under the conditions of nonlinear motion and shielding of a target is enhanced, and the stability and continuity of tracking are improved; a fusion multi-stage matching strategy is adopted to effectively improve the matching accuracy and robustness in a complex environment; meanwhile, a confidence shunt strategy and a trajectory management mechanism are combined, so that the response speed to a new target and the fault-tolerant capability to a lost target are improved; through combination of global motion estimation and height information judgment, accurate correction and target distinguishing of the track of the unmanned aerial vehicle are realized, aliasing and misjudgment are avoided, and the detection and tracking capability of the unmanned aerial vehicle is remarkably improved.

The invention relates to the field of digital image processing, in particular to a super-resolution imaging method based on focal plane splicing and adaptive fusion. According to the method, sub-pixel offset among nine CCDs is preset through hardware, and nine frames of low-resolution image sequences with accurate displacement are obtained in push-broom. A central image is taken as a reference frame, high-precision mapping is realized based on hardware offset, motion estimation errors are avoided, effective pixels are screened by calculating robustness weight, an anisotropic Gaussian kernel function with a self-adaptive local structure is constructed so as to maintain image edge and detail features, and each frame is accumulated to a high-resolution grid in a weighting mode, so that a high-resolution image is obtained. And a sample compensation mechanism based on cumulative robustness is introduced, a fusion strategy is adaptively adjusted in an information insufficient area, and finally a high-resolution image is generated through normalization. The method significantly improves the imaging quality, suppresses artifacts and noise, and is suitable for the field of satellite remote sensing.

The invention discloses an ultra-high-definition video stream adaptive coding method based on deep learning visual saliency, and the method comprises the steps: carrying out the five-scale Gaussian filtering processing and image pyramid construction of a video frame, and combining Sobel gradient, Laplacian edge and local binary pattern feature extraction to generate a multi-scale feature map; a pre-training saliency detection network is adopted, and a smooth saliency thermodynamic diagram is generated through processing of a feature adaptation layer, a residual encoder, a self-attention mechanism and a transposed convolution decoder; dividing the video frame into a high region, a middle region and a low region according to the saliency thermodynamic diagram, and establishing a regionalization coding parameter table; performing differentiated prediction modes, motion estimation and quantization strategies on different salient regions; and organizing coded data according to an H.265 / HEVC standard, and embedding the saliency thermodynamic diagram into supplementary enhancement information for transmission. According to the method, the important region concerned by the user can be intelligently identified, a differentiated coding strategy based on content semantics is realized, and the coding efficiency is remarkably improved.

The invention discloses a multi-view target detection tracking method and device based on adaptive fusion and time sequence association, and relates to the technical field of visual tracking, and the method comprises the steps: synchronously obtaining the input images of the current frames of a plurality of camera devices, and carrying out the multi-scale feature extraction and adaptive multi-scale feature fusion, determining scale fusion features of the input images; mapping each scale fusion feature to a bird's-eye view space, and splicing to generate current bird's-eye view information; inputting the current aerial view information and the historical aerial view information of the adjacent historical frames into an improved aerial view time sequence Transform to output aerial view features; after the aerial view features are decoded, aerial view detection and tracking results are determined through a detection head and a tracking head. Through adaptive multi-scale feature fusion, artifacts are reduced, feature detail expression is enhanced at the same time, a bird's-eye view sequence Transform is adopted to perform pixel-level matching between adjacent frames under a bird's-eye view angle so as to enhance fine-grained motion estimation, and the detection and tracking accuracy can be improved.

The invention relates to the technical field of steel wire rope state detection, in particular to a steel wire rope diameter measuring method and system based on event compensation correction. According to the technical scheme, the system comprises an imaging module, a boundary initial extraction module, a dynamic motion estimation module, a dynamic correction compensation module and a diameter calculation module. The imaging module comprises a traditional camera, an event camera, an optical component, a synchronous triggering device and a spectroscope. The traditional camera is used for collecting frame images of the steel wire rope. According to the method, on the basis of complementary advantages of an event camera and a traditional camera, systematic measurement deviation under the dynamic working condition is solved from the mechanism through an event-driven dynamic geometric compensation model, and quasi-static imaging dependence is broken through. The method does not need extra constraint or increase of frame rate, control cost and hardware complexity, avoids the problems of reduction of signal-to-noise ratio and the like, realizes high-precision stable measurement under high-speed and vibration working conditions, adapts to online monitoring of multiple industrial scenes, and is remarkable in practicability and application value.

The present invention discloses a key joint motion estimation based method for estimating continuous human postures. A motion estimation block matching algorithm is applied to human key joint tracking, so as to obtain continuous human posture results. Meanwhile, the results are continuously corrected by using a deep neural network based human posture estimator. The present invention may estimate the continuous human postures in a video stream, where the human postures are specifically embodied as coordinate positions of human joints in a video frame. Compared with a posture estimation method completely relying on a deep neural network, the posture estimation method provided by the present invention has the advantages of high frame rate, low hardware requirements, and sequential continuity of recognition results; and compared with a posture estimation method completely relying on a motion estimation algorithm, the present invention may correct a cumulative error, to improve the estimation accuracy.

The invention provides a power grid photovoltaic output short-term prediction method and system fused with a meteorological time sequence, and relates to the technical field of photovoltaic power, and the method comprises the steps: reading multi-source meteorological observation data of a target region; performing cloud field motion estimation, and establishing a cloud motion vector field; activating a temperature corrector, reading environment temperature and photovoltaic module temperature, constructing a temperature-module efficiency dynamic model by utilizing a temperature reading result and wind speed and wind direction data of a weather station, and establishing a power upper limit correction factor; after an illumination disturbance signal is established by using the cloud motion vector field, a short-time scale dynamic output prediction curve is generated; and performing power grid operation constraint mapping adjustment, and outputting a calibration prediction result. According to the method and the device, the technical problem of low photovoltaic output short-term prediction precision caused by instability and unpredictability due to the fact that photovoltaic power generation is easily influenced by weather changes in the prior art is solved, and the photovoltaic output prediction precision is improved and the power grid dispatching and operation are optimized by fusing the meteorological time sequence.

Aspects presented herein relate to methods and devices for display processing including an apparatus, e.g., a GPU or a CPU. The apparatus may obtain a first frame and a second frame of a plurality of frames in a scene. The apparatus may also calculate a set of first motion vectors in the first frame and a set of second motion vectors in the second frame, where the set of first motion vectors and the set of second motion vectors are calculated based on a block matching procedure. Further, the apparatus may estimate at least one third frame in the plurality of frames based on the set of first motion vectors in the first frame and the set of second motion vectors in the second frame, where the at least one third frame is subsequent to the second frame in the plurality of frames.

Systems and techniques are described herein for interpolating image data. For instance, a method for interpolating image data is provided. The method may include processing a first image frame and a second image frame using a motion estimator to generate first motion vectors, wherein the motion estimator comprises a machine-learning model trained to generate motion vectors based on image frames; projecting the first motion vectors to generate second motion vectors; and generating a third image frame based on the first image frame, the second image frame, and the second motion vectors.

The invention discloses an anti-jitter data acquisition and obstacle recognition method for a rail transit vehicle, and relates to the technical field of rail transit vehicle data acquisition and processing, continuous four frames of original images are acquired through a binocular camera by adopting synchronous and staggered shooting, and global motion estimation is performed in combination with acquired sensor data; local pixel level alignment is realized by using a preset image layer and pixel grid division, a stable image is generated through weighted fusion, local features of the image are optimized based on a multi-head attention mechanism, the stable image is mapped to a high-dimensional vector space by using a depth encoder, motion correction information is fused, and finally, the motion correction information is obtained. A deep reconstruction network is combined with binocular parallax to generate a three-dimensional point cloud, three-dimensional coordinate changes are monitored in real time, obstacle recognition and early warning are achieved, high-quality images can be obtained in the violent vibration environment of a vehicle, the data acquisition precision and the real-time performance and accuracy of obstacle detection are improved, and the operation safety of rail traffic is effectively guaranteed.

Systems and methods are disclosed that perform global human and camera motion estimation using a motion diffusion model that is attached to a control branch. For instance, using a controlled motion denoiser that comprises the motion diffusion model and the control branch, global human motions and the corresponding camera motions from “in-the-wild” videos may be estimated. Initially, SLAM may be used to initialize the camera motion and a pose estimation model may be used to estimate the local human motion. Combining the two, embodiments of the present disclosure initialize the global human motion. Then, during optimization and using a COIN system that includes the controlled motion denoiser and / or using a COIN algorithm, embodiments of the present disclosure enforce the global human and camera motion to satisfy a two-dimensional (2D) projection on videos and the motion distribution from the motion diffusion model.

A method for scale-aware depth estimation using multi-camera projection loss is described. The method includes determining a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes training a scale-aware depth estimation model and an ego-motion estimation model according to the multi-camera photometric loss. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the scale-aware depth estimation model and the ego-motion estimation model. The method also includes planning a vehicle control action of the ego vehicle according to the 360° point cloud of the scene surrounding the ego vehicle.

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.

Learning extraction of movement information from sensor data includes providing a time series of frames of sensor data recorded by physical observation of an object, providing a time series of object boundary boxes each encompassing the object in sensor data frames, supplying the object boundary box at a time t, as well as a history of sensor data from the sensor data time series, and / or a history of object boundary boxes from the time series of object boundary boxes, prior to time t to a trainable machine learning model which predicts an object boundary box for a time t+k, comparing the predicted object boundary box with a comparison box obtained from the time series of object boundary boxes for the time t+k, evaluating a deviation between the predicted object boundary box and the comparison box using a predetermined cost function, and optimizing parameters which characterize the behavior of the model.

The invention discloses a ViT model lightweight method for video action recognition, which comprises the following steps of: performing motion estimation in a data processing stage, calculating to obtain motion intensity and time redundancy scores of tokens, and performing window division on video frames according to the scores. Afterwards, a window-global token merging strategy is adopted to effectively merge redundant tokens, the first three layers of the model process local information through adaptive window merging, and from the fourth layer, the model repeatedly executes global merging according to the depth of the layers so as to realize gradually enhanced feature representation; therefore, the number of tokens is reduced, the complexity of the ViT model in space-time self-attention calculation is reduced, and the problem of high calculation resource consumption caused by space-time information redundancy of the current ViT model in a video task is solved. Calculation overhead and energy consumption can be remarkably reduced in tasks such as video action recognition, action detection and time sequence event understanding, and meanwhile the characterization capacity of a key dynamic area is kept.

The invention relates to an end-to-end optimized prediction video coding method and device, and belongs to the field of video coding. According to the method, a B frame coding model performs motion estimation on a to-be-coded frame by respectively utilizing information of a forward reference frame and a backward reference frame, so that motion information and content change in a video sequence can be more accurately captured, more accurate characteristics and prior information are provided for residual coding, and the coding performance of the model is further improved. Through the bidirectional information acquisition and processing mode, the comprehensive understanding of the time-space characteristics of the video sequence is improved, and the processing of the model on the motion and content change in the coding process is effectively optimized, so that the more excellent video compression performance is realized.

Described herein are systems, methods, and instrumentalities associated with estimating the motions of multiple 3D points in a scene and predicting a view of scene based on the estimated motions. The tasks may be accomplished using one or more machine-learning (ML) models. A first ML model may be used to predict motion-embedding features for a temporal state of a scene, based on motion-embedding features for previous states. A second ML model may be used to predict a motion field representing displacement or deformation of the multiple 3D points from a source time to a target time. Then, a third ML model may be used to predict respective image properties of the 3D points based on their updated locations at the target time and / or a viewing direction. An image of the scene at the target time may then be generated based on the predicted image properties of the 3D points.

The invention discloses a spatial moving target optical comb rapid distance measurement method and system based on a hierarchical guidance strategy, relates to the technical field of spatial moving target distance measurement, and aims to solve the problems of long alignment time, high energy consumption and high efficiency caused by the fact that traditional optical alignment before moving target scanning depends on a preset global scanning mode. And the requirements of dynamic tracking distance measurement are difficult to meet. The system is divided into a laser emitting and scanning part, an image receiving and processing part and a processor part. The laser emission scanning part utilizes the deflection of a fast reflecting mirror to change a laser light path, and scans a reflecting prism on the surface of a target; the image receiving and processing part reflects an image of a target reflecting prism to a binocular camera through a small fast reflecting mirror, and then inputs an imaging result of the camera into an image processing module to calculate a relative pose with a target; and the main processor performs target motion estimation and scanning path planning according to the relative pose information, and inputs reference path information to the fast steering mirror driving circuit to implement tracking control. Through a multi-sensor fusion sensing algorithm, accurate target sensing and positioning in a complex space environment are realized, interference such as in-orbit platform vibration and space illumination condition change under a microgravity condition can be effectively coped with, and thus the robustness and accuracy of space target detection are greatly improved.

The invention provides a curvature-enhanced large-displacement image variational optical flow method for motion estimation, relates to the field of image processing, and aims to describe the local structure complexity of an image by introducing an image contour curvature. On the basis of the curvature, limited self-adaptive weighted adjustment is carried out on the brightness invariant constraint and the gradient invariant constraint on the data item level of the opto-rheological model, so that the interference of unreliable matching in a complex structure region on optical flow estimation is inhibited; the robustness of optical flow estimation in illumination variation, complex texture and large displacement scenes is improved; and the numerical stability and convergence of the model in the multi-scale calculation process are ensured.

Systems and methods herein are for a video encoder to be associated with a temporal filter and a coding tree and that can perform a main pass for video encoding using individual video blocks towards prediction of at least one frame associated with the media stream, where the coding tree is associated with a lookahead pass, and where the temporal filter can enable denoising within the lookahead pass to reduce an effect of noise in one or more of motion estimation or mode selection of the video encoding.

Systems and methods for reducing latency through motion estimation and compensation techniques are disclosed. The systems and methods include a client device that uses transmitted lookup tables from a remote server to match user input to motion vectors, and tag and sum those motion vectors. When a remote server transmits encoded video frames to the client, the client decodes those video frames and applies the summed motion vectors to the decoded frames to estimate motion in those frames. The server instructs the client to receive input from a user, and use that input to match to cached motion vectors or invalidators. Based on that comparison, the client then applies the matched motion vectors or invalidators to effect motion compensation in a graphic interface. In this manner, latency in video data streams is reduced.