129 results about "Motion estimation" patented technology

The invention discloses a real-time single-stage remote sensing image correction target detection method based on YOLOV8, and relates to the technical field of remote sensing image processing. According to the method, a deformable convolution dynamic prediction local geometric distortion parameter is embedded based on a YOLOv8 backbone network, an adaptive deformation field is generated, pixel-level real-time correction is realized, shallow details and high-level semantic features are fused through a bidirectional path aggregation network, and channel attention and a space gating mechanism are combined, so that the real-time correction of the image is realized. The small target detection capability is enhanced, background noise is suppressed, angle prediction is divided into discrete classification and continuous residual error regression tasks through a decoupling type rotation detection head, angle periodic errors are eliminated in combination with a direction sensitive loss function, and the rotation frame positioning precision is improved. And constructing a dynamic multi-task collaborative loss function, introducing gradient distribution consistency constraint to jointly optimize correction and detection tasks, and realizing feature semantic alignment and model self-enhancement through end-to-end closed-loop training. And the rotating target detection precision and the complex scene robustness are obviously improved.

The invention discloses a visual inertial positioning method based on dynamic target detection and semantic information constraint, and belongs to the field of motion estimation and dynamic environment processing. According to the method, a dynamic target detection mechanism is introduced, the dynamic target is effectively detected based on the target detection network, inertial navigation information and geometric constraints, the dynamic target is effectively recognized in the image processing process, the corresponding dynamic feature points are screened out, the mismatching rate of the dynamic features is remarkably reduced, and high-quality observation input is provided for back-end optimization. In the back-end sliding window optimization stage, a semantic information consistency constraint method is constructed, and the estimation stability of the system in a weak texture area or a repeated texture area is enhanced by utilizing the consistency of feature points in the same semantic area on a geometric structure. Visual inertia pose estimation is realized based on dynamic target detection and semantic information constraint, and high-robustness and high-precision pose estimation can still be realized in a complex environment with dynamic interference of pedestrians, vehicles and the like and severe scene change.

The invention discloses a mining equipment positioning system and method based on radar, ultra wide band and inertial navigation, and the system is characterized in that a multi-source sensor unit, a data synchronization and collection module and a preprocessing module are connected in sequence, and a state fusion module is connected with the preprocessing module, a UWB ranging error compensation module, a radar motion estimation module and a pose output module; the upper computer is connected with the pose output module; the method comprises the steps of collecting multi-source data; timestamp alignment and format conversion processing are carried out, and feature and speed constraint information extraction is carried out on the point cloud data; fitting correction is carried out on the ranging data by adopting a weighted least square method, and position information is obtained through an error weighted average algorithm; feature matching is carried out, and in combination with a KISS-ICP point cloud registration method, a speed estimation result is obtained; carrying out data fusion, and obtaining a pose estimation result by adopting an extended Kalman filtering algorithm; and outputting equipment position and attitude information. According to the invention, high-precision real-time estimation of the attitude and position information of the mining equipment can be realized.

The invention relates to an end-to-end automatic driving method based on Transform multi-modal feature fusion, and the method comprises the steps: collecting an RGB image and an original depth image, and converting the original depth image into an HHA image; inputting the RGB image and the HHA image into a sensing module for fusion to obtain a fusion feature map, and performing global average pooling and flattening operation to obtain environment features; acquiring the real-time speed of a vehicle, an advanced navigation command and a target position, connecting in series to form measurement input, and processing based on an MLP measurement encoder to obtain measurement characteristics; adding the environment features and the measurement features element by element to obtain combined features, performing downsampling step by step, and connecting a ReLU activation function behind each layer to obtain track features; and a loss estimator is constructed, training loss is dynamically predicted in real time based on the track features, and the weight of each feature parameter is dynamically adjusted according to the loss, so that the vehicle is dynamically controlled and adjusted in real time. According to the method, the multi-modal features can be well fused, so that a more accurate end-to-end automatic driving method is realized.

The invention relates to the technical field of advertisement shooting, in particular to a holder self-adaptive tracking control method based on visual identification, which comprises the following steps of: acquiring a scene image digital signal in real time through a camera, then extracting the coordinate, the size and the motion trail of a detection target, and adopting an optical flow motion estimation algorithm in the extraction process, the optical flow motion estimation algorithm can calculate the motion information of the target by analyzing the motion of the pixel points in the image, and is combined with pure extraction of static information such as coordinates and sizes, so that the state of the detected target can be described more comprehensively and dynamically, the location of the target can be known, and the motion of the target can be accurately mastered; and then, according to the motion trail of the detection target, calculating and establishing a mapping relation between a pan-tilt angle and a camera visual field, and calculating the angle of the pan-tilt needing to be adjusted when the detection target shoots the advertisement and the detection target moves through the mapping relation between the pan-tilt angle and the camera visual field, so that the detection target can be always located at the central point coordinate of the image digital signal.

The invention relates to the technical field of machine learning, and particularly discloses an intelligent parking guidance and reverse vehicle searching system based on multi-source heterogeneous data fusion. The system comprises a data perception fusion layer, a dynamic prediction decision-making layer and a user service interaction layer, realizes multi-step advanced probability prediction and dynamic optimal path planning of a parking space state through fusion of geomagnetic detection, video identification, payment flow, traffic situation and activity information, and realizes the optimal path planning of the parking space state based on multi-mode induction and live-action AR reverse vehicle searching guidance. And the parking efficiency and the user experience are improved.

The invention belongs to the technical field of visual angle follow-up vehicle panoramic look-around, and particularly relates to a visual angle follow-up vehicle panoramic look-around system based on dynamic prediction, which is characterized in that monocular vision and laser radar time sequence data are fused, and a dense and stable dynamic target depth sequence is generated through dynamic target initial perception, cross-modal time sequence alignment and inter-frame depth correlation modeling; through environment interaction modeling and target behavior intention prediction double-branch parallel processing, a dynamic self-adaptive accurate track is output in combination with a weighted fusion strategy, and the prediction weight dynamic adjustment requirement of a complex motion scene is adapted; after a virtual view angle image is preliminarily synthesized based on the predicted trajectory, deviation detection and coordinate dynamic correction are realized through inter-frame target consistency verification, and artifact area positioning and pixel-level restoration are synchronously completed; multi-modal preprocessing and algorithm tasks are deployed according to hardware, and a model lightweight optimization and inter-frame prediction result multiplexing strategy is combined, so that low-delay operation of the system in a high-resolution panoramic image output scene is ensured.

The invention discloses a cracking furnace intelligent optimization system control method and system architecture, and belongs to the cracking furnace application field, and the cracking furnace intelligent optimization system control method comprises the following steps: S1, collecting cracking furnace operation time sequence characteristic data and flame combustion video data in real time through a sensor and a high temperature camera; s2, preprocessing the collected multi-modal data, S3, integrating the preprocessed multi-modal data by using federal learning, and constructing a cross-modal causal logic model; s4, designing a deep reinforcement learning framework, introducing the dynamic prediction model of the cracking furnace, and reasoning the recommendation range of each parameter of the optimized operation of the cracking furnace; s5, dynamic process data and the environment-friendly emission prediction model are combined, a data-driven MPC algorithm framework is adopted, and dynamic optimization control over the cracking furnace is achieved; and S6, the operator performs adjustment or selects emergency operation according to the recommended parameters and strategies to realize man-machine cooperative control. Finally, the dynamic optimization control of the cracking furnace can be realized, and the energy consumption is effectively reduced.

The invention relates to the technical field of security and protection monitoring systems, in particular to a full-scene intelligent analysis system and method based on a multi-modal large model, and the system comprises a feature extraction module which is used for obtaining cross-modal data, generating a time sequence feature set according to a cross-modal tensor, and generating a space structure set according to cross-modal image data; the semantic fusion module performs semantic interaction on the time sequence feature set and the space structure set based on a cross-modal structure to obtain a cross-modal data set; the causal generation module generates causal association based on the time sequence feature set and the spatial structure set; generating a causal time sequence association based on the cross-modal data set and the causal association; the dynamic prediction module associates the dynamic learning association mechanism based on the causal timing to generate a dynamic prediction model. According to the invention, the intelligent supervision capability of key places is obviously improved, and the generalization detection capability of novel violation behaviors can be improved, so that complex or unknown risk scenes are covered.

The invention relates to the technical field of intelligent equipment scheduling planning, and discloses a scene scheduling method and system for intelligent equipment, and the method comprises the steps: determining the task emergency degree and the motion complexity based on a dual-domain scheduling factor construction mechanism; constructing a scheduling priority weight; screening a conflict node set; generating an avoidance decision by adopting a double-case evaluation mechanism; and finally, synchronously issuing the avoidance decision to related intelligent equipment. In the prior art, there is no scheduling method of a priority guidance and fine conflict judgment mechanism, and especially in a dense operation scene of multiple mobile devices including an industrial AGV, a luggage transfer robot and a mobile charging vehicle, dynamic prediction and active avoidance of conflict risks are difficult to realize. Due to the fact that task and motion double-domain factor construction and a space and time double-condition conflict recognition mechanism are introduced, active avoiding and real-time response of device-level scheduling are achieved, and the cooperative operation stability of intelligent devices is improved.

The invention discloses a mechanical arm intelligent grabbing system and method based on three-dimensional vision. The system comprises a three-dimensional vision system, an image preprocessing module, a model matching module, a grabbing point generation module, a grabbing control module and a dynamic environment self-adaption module. The method comprises the steps of multi-modal reference model collection and feature library establishment, precise hand-eye calibration and grabbing strategy self-learning and multi-modal fusion recognition, and dynamic prediction and self-adaptive grabbing. By adopting the three-dimensional visual recognition system and combining algorithms such as point cloud processing, point cloud method phase generation, point cloud downsampling, rough model matching and fine model matching, the limitation that a traditional mechanical arm system depends on two-dimensional image training and position coordinate point matching is overcome; by generating the grabbing points and calculating the pose coordinates of the grabbing points according to the hand-eye calibration matrix, the problem that the estimated pose of the grabbed object under the camera coordinate system is inaccurate in the prior art is solved. The system is easy to implement and has high practicability and popularization value.

The invention belongs to the technical field of artificial intelligence, and discloses a hidden space world model construction method for a robot grabbing operation scene and related equipment thereof. The method comprises the steps of performing semantic segmentation on a scene image through an image semantic segmentation network to generate a semantic segmentation mask of each object; processing the semantic segmentation mask through a multi-task scene understanding network to generate implicit representation; based on the implicit representation of each object at the current moment and the action information of the mechanical arm, predicting implicit representation at the next moment through a state transition network; and according to the predicted implicit representation at the next moment, generating a scene segmentation image reconstruction result, an object existence judgment result and a contact relation judgment result through a decoder in the multi-task scene understanding network architecture, and constructing an implicit space world model of the scene. Based on the method, efficient characterization and dynamic prediction of the robot grabbing operation scene are achieved, and the generalization ability and the multi-task cooperation efficiency of the model in a complex environment are improved.

The invention discloses a three-dimensional motion estimation method based on an event camera. The method comprises the following steps: 1, constructing a time-space decoupling event time sequence and spatial representation; 2, geometric structure and context features are extracted through a space encoder, and time sequence evolution features are extracted through a time sequence encoder; 3, respectively constructing a spatial correlation graph and a time sequence correlation graph based on the geometric structure characteristics and the time sequence evolution characteristics; 4, constructing a pixel track curve according to the event density; 5, geometric structure and time sequence evolution characteristics are sampled along the pixel track curve, and a space-time correlation graph is generated; 6, iteratively optimizing pixel track curve parameters according to the space-time correlation graph and the context features; and 7, utilizing the optimized pixel track curve and the derivative thereof to accurately predict the three-dimensional motion. The method obviously improves the three-dimensional motion estimation precision of the event camera in a complex scene, is suitable for the fields of automatic driving and robots, and has a good engineering application value.

The invention provides an HRRP target micro-motion characteristic extraction method based on chaos optimization and polarization synthesis, and aims at the field of high-resolution one-dimensional range profiles. The method comprises the following steps: performing optimal polarization contrast enhancement and optimal polarization fusion based on a maximum signal-to-noise ratio criterion on HRRP multi-channel polarization data; performing period estimation on the polarization fusion data in combination with second harmonic back-off optimization and autocorrelation / amplitude difference enhancement processing; chaotic mapping and back propagation are combined to avoid the problem that traditional optimization is prone to falling into local minimum, and precession angle and size results of the space target are obtained. According to the method, the feature contrast of HRRP data is remarkably improved through polarization fusion, meanwhile, high-precision estimation of the micro-motion features of the space target is achieved under the condition of a low signal-to-noise ratio, the stability of periodic detection within limited observation time and the global search capability of parameter inversion are effectively improved, and the method is suitable for large-scale popularization and application. And the accuracy and robustness of space target micro-motion estimation in a complex environment are obviously enhanced.

The invention discloses a water conservancy and hydropower potential safety hazard intelligent identification method and system based on image identification, and relates to the technical field of water conservancy and hydropower engineering management. The method comprises the following steps: acquiring a real-time image data sequence and an engineering facility digital twinborn model; performing space mapping and timestamp alignment on each frame of image and the digital twin model to generate a registered time-space image data set; layered feature extraction is carried out on the registered time-space image data set, and structured engineering defect features are analyzed from potential abnormal targets in combination with three-dimensional geometric structure information of a digital twin model; performing association analysis on the engineering defect features and engineering attribute information and historical operation and maintenance data of the digital twin model to judge hidden danger types of the engineering defects, and calculating risk levels of the engineering defects according to a preset risk assessment matrix; and generating a structured potential safety hazard report. The method realizes high-precision context sensing of potential safety hazards of water conservancy and hydropower and has dynamic prediction capability.

The invention relates to a visual positioning method and device based on a panoramic camera, equipment and a storage medium. The method comprises the following steps: acquiring a panoramic image through a double-fisheye camera; performing SPHORB feature point extraction on the panoramic image, and matching feature points between two continuous frames of panoramic images by using an optimized feature matching algorithm to obtain matched feature point pairs; based on the feature point pairs, calculating relative motion of the double-fisheye camera between two continuous frames of panoramic images by using a motion estimation algorithm to obtain a pose transformation matrix of the double-fisheye camera; and optimizing the pose transformation matrix by using a bundle adjustment optimization algorithm to obtain an optimized pose transformation matrix. According to the embodiment of the invention, the double-fisheye camera is used for shooting the panoramic image, the visual range of the camera is expanded, more comprehensive environment information is obtained, feature point extraction is carried out on the panoramic image, and the visual positioning precision of equipment is improved.

The invention relates to the field of intelligent medical treatment, in particular to a dynamic illumination prediction method and device based on a refraction state and a program product. Comprising the steps of obtaining eye refraction data of a to-be-tested person; carrying out clustering and feature fusion on the refractive data to obtain fusion features; outputting the fusion features to a trained prediction model for prediction to obtain a brightness prediction value and a color temperature prediction value; and the prediction model carries out weight parameter initialization by compressing an abnormal value of refractive data and compensating distribution deviation of refractive clustering. According to the method, personalized expression of different myopia degrees on illumination perception differences can be achieved, brightness and color temperature parameters are dynamically predicted, actual lighting equipment is driven, dynamic balance of visual comfort and environmental adaptability is achieved, and the method has good actual application value.

A system including an internal sensor configured to generate first sensor data, an external perception sensor configured to generate second sensor data, at least one memory configured to store computer executable instructions, and at least one processor is disclosed. The at least one processor is coupled to the internal sensor, the external perception sensor, and the at least one memory. The at least one processor is configured to execute the computer executable instructions to: (i) compute a first velocity output based upon the first sensor data; (ii) compute a second velocity output based upon the second sensor data; (iii) compute a relative velocity between the first velocity output and the second velocity output; (iv) determine the relative velocity is outside a predetermined threshold value; and (v) perform an intervening action in response to the determining.

The application belongs to the technical field of computer vision and image processing. The application provides a large field of view underwater camera calibration-free distortion correction method. The disclosure embodiment can adopt a two-stage motion estimation architecture of "region-level progressive TPS plus pixel-level residual flow compensation", decouple different complexity motion representations, and balance global deformation fitting and local detail preservation; a lightweight task classifier is designed to identify distortion types based on TPS control point distribution features; a prompt learning module is introduced to modulate decoder features through learnable prompt words and task probability vectors to achieve single-model multi-task adaptive correction; a multi-task joint training strategy and a hybrid loss function are used to strengthen the generalization ability of the model to different distortion types and support arbitrary resolution image input.

The application belongs to the technical field of image recognition, and provides a target tracking method and system based on AI vision, which comprises the following steps: original video frame acquisition; environment adaptive image enhancement; multi-target detection and multi-modal feature extraction; local optical flow motion estimation; multi-target trajectory association; occlusion processing and re-identification; trajectory output and result analysis; the method sets a physical-depth learning cascade defogging model, dynamically switches light / thick fog processing paths through dark channel mean, fuses atmospheric scattering physical priori and U-Net residual correction, has an environment adaptive perception architecture, breaks through the failure bottleneck of traditional single model under sudden fog concentration, constructs a trajectory cognition system coupled with appearance-motion-geometry, designs a dynamic cost matrix and a feature cache pool, solves the ID switching problem caused by similar target aggregation and long-time occlusion, and improves the accuracy of target tracking in an occlusion environment.

The present invention discloses a 3D biomedical video compression method, system, device and storage medium. Through a 3D biomedical video compression framework including 3D motion estimation, motion vector coding, 3D motion compensation and residual coding, the 3D biomedical video compression performance can be greatly improved and important information in the 3D biomedical video can be retained.

The invention discloses a photoelectric turret visual axis stability control method based on self-adaptive disturbance prediction compensation and a computer product, and the method comprises the following steps: S1, establishing a gyro speed loop discrete model, calculating the deviation between the output of the model and the angular velocity of a visual axis measured by a gyro, and obtaining a disturbance estimation value through an inverse model and a low-pass filter; s2, constructing a mathematical model of disturbance based on a recursive least square method, and performing multi-step forward prediction on equivalent disturbance by using the model; and S3, carrying out weighted fusion on the predicted equivalent disturbance and the real-time disturbance estimation value, and finally feeding a fused compensation signal forward to the input end of the control system to counteract the influence of the actual disturbance on the system. According to the method, real-time estimation and dynamic prediction are fused, the learning-prediction capacity is added on the basis of perception-compensation of a traditional DOB, the method is suitable for a photoelectric turret control system which is remarkable in periodic disturbance or sensitive to delay, phase lag of the traditional DOB is reduced, and the stable control bandwidth and disturbance rejection capacity of a visual axis are greatly improved.

The embodiment of the invention provides a data processing method and device and track capturing equipment, and the method comprises the steps: firstly obtaining magnetic data and video frame data in a motion process, and then obtaining a motion track of a motion object in the motion process according to the magnetic data and the video frame data. That is to say, the motion trail of the moving object is obtained through the magnetic data obtained through the magnetic positioning technology and the video frame data obtained through the video collection device. The method does not need to depend on experience and judgment of observers, and is not easily interfered by external factors. In addition, the method can make full use of the high-frequency sampled magnetic data of the magnetic positioning technology to dynamically predict and compensate the low-frame-rate video frame data, and can improve the positioning precision of the motion trail of the moving object compared with the method of obtaining the motion trail of the moving object by using single video frame data.

Various embodiments include techniques for encoding media frames using a multistage search without the processing overhead of pyramidal motion estimation techniques. The first stage of the multistage search generates many motion vectors, where each motion vector is based on a small range full pixel search of a pixel group in the media frame. The video encoder selects a full pixel best motion (FBM) vector for all pixel groups from the motion vectors generated during the first stage. The second and subsequent phases of the multistage search is based on the FBM vector from the prior stage as a starting point, where the search range for each subsequent phase is larger than the search range of the prior stage. The multistage search can be performed over a fixed number of stages. Alternatively, the multistage search can be terminated when the cost value for the current stage is below a threshold value.

The embodiment of the invention discloses a deblurred view synthesis method and device, electronic equipment and a storage medium, and relates to the technical field of computer vision and three-dimensional reconstruction, and the method comprises the steps: obtaining an initial camera pose through three-dimensional reconstruction software, and employing a cubic Bezier curve to model a camera motion track; dynamically predicting three-plane resolution by using a feedforward neural network according to the track length; self-adaptive three-plane downsampling is carried out based on the predicted resolution, and a clear image is generated through a volume rendering technology; and finally, obtaining a full-resolution three-plane field through a loss optimization and gradient descent algorithm, and generating a clear image corresponding to the pose of the target camera according to the learnable parameters and the three-plane field. According to the method, the problems of low sampling efficiency, long training time and rendering quality degradation under a long track when facing camera motion tracks with different lengths in the prior art are solved, the training efficiency is remarkably improved, the memory occupation is reduced, and the generalization ability and reconstruction quality of the model are enhanced.

The invention provides a science and technology project performance dynamic prediction method and system combining digital twinning and artificial intelligence, and relates to the technical field of science and technology project management data analysis. According to the method, a performance indicator system and project process data are obtained, a mapping relation between indicator calibers and process data is established, a project digital twinborn body containing task dependence, milestone and resource boundary constraints is constructed, and a feature sequence with consistent constraints is generated; on the basis, a consistency constraint training performance prediction model is introduced, rolling prediction and confidence evaluation in the operation period are achieved, calibration of the model or the mapping relation is triggered when prediction credibility is insufficient or management constraint is violated, and therefore the caliber consistency, constraint rationality and trend availability of a performance prediction result are improved.

The invention discloses a synchronous control system for multi-channel small lamp sequential light emitting and three-line array camera synchronous shooting and a time delay compensation method thereof, and the system comprises a high-performance control module, a time delay detection module, a small lamp driving module, a camera triggering module, a parameter storage module and a man-machine interaction module. According to the time delay compensation method, multi-scale delay modeling, dynamic prediction and layered closed-loop control principles are combined, a real-time time delay characteristic function is established by using exponential weighted filtering and an autoregression prediction model, error self-correction is realized through proportional-integral adaptive control, and an outlier rejection mechanism is introduced to suppress noise interference. The system can be applied to scenes such as industrial visual detection, scientific experiment high-speed imaging, medical visual acquisition and motion analysis, the problem of synchronization errors caused by delay drift and signal coupling in a multi-channel system is effectively solved, stable synchronization of the system under the high-frequency triggering condition is achieved, and the errors are controlled within + / -5 microseconds.

The invention belongs to the technical field of construction progress early warning, and particularly relates to a project construction progress early warning method, system and equipment based on deep learning, and the method comprises the steps: carrying out the image recognition of a construction scene image based on an improved R-FCN algorithm, obtaining image recognition data, enabling the image recognition data and quality monitoring index data to form multi-modal input data, and carrying out the construction progress early warning. Inputting the data into an LSTM deep learning model to predict future quality monitoring index data; and outputting a risk membership degree and a D-S evidence theory for predicted future quality monitoring index data through a cloud model to obtain a fusion result, weighting the risk membership degree and the fusion result to generate a comprehensive risk index, and realizing construction progress grading early warning according to the comprehensive risk index. According to the method, on one hand, the dynamic prediction precision of the quality monitoring index data can be improved through the multi-modal input data, the method can dynamically adapt to complex construction scenes, on the other hand, conversion from quantification to qualitative is achieved through the comprehensive risk index, and finally construction progress grading early warning is achieved.

The video super-resolution system comprises a motion estimation device, a mapping device and a neural network super-resolution device. The mobile estimation device calculates an optical flow according to the current frame and a previous frame received from the memory. The mapping device maps a previous frame and a previous output received from the memory according to the optical flow to generate a mapped frame and a mapped output, respectively. The neural network super-resolution device performs feature extraction on the current frame, the mapping frame, the mapping output and the count value to generate at least one feature, and performs deep learning on the at least one feature and a previous hidden state of the previous output to generate a current hidden state and a deep learning result, and performing feature extraction on the deep learning result to generate a current output. The neural network super-resolution device stores the current frame, the current hidden state, and the current output to a memory.

The application relates to the technical field of satellite meteorological prediction, and discloses an optical flow-depth learning satellite multi-channel data extrapolation method. The method comprises the following steps: acquiring and preprocessing time-series satellite multi-channel data; a two-dimensional motion field is estimated by using a variational optical flow algorithm, preliminary space-time extrapolation is carried out in combination with a semi-Lagrangian algorithm and a mass conservation constraint; a deep learning model integrating space-time convolution and a recurrent neural network is constructed, and the preliminary extrapolation result and original deep features are jointly trained and predicted; two types of results are dynamically fused through an adaptive weight fusion mechanism; finally, the fusion result is corrected based on atmospheric dynamics and radiation transmission physical constraints, and a physically consistent multi-channel extrapolation field is generated. The method combines the advantages of physical motion estimation and data-driven learning, and improves the accuracy and rationality of satellite observation extrapolation from the minute level to the hour level.