20 results about "Motion vision" patented technology

The invention discloses a motion visual evaluation system and method combining attitude motion and visual training, and relates to the technical field of digital medical treatment and visual health, and the system comprises a biomechanics-attitude acquisition module which integrates a three-dimensional force sensor, a high-precision joint angle sensor, an inertial measurement unit sensor group and a depth camera; synchronously collecting joint torque, ground reaction force, muscle power output, joint angle and posture offset; through a mechanics-vision correlation model of the data processing module, how the biomechanical characteristics influence the visual response and the feedback effect of the visual task on the biomechanical state can be deeply excavated, the improvement of the posture and the visual ability is concerned, the training strategy can be optimized from the biomechanical level, and the training efficiency is improved. And a key basis is provided for the evaluation feedback module to output a multi-dimensional report containing collaborative rating, biomechanical optimization direction and visual adaptation suggestions.

The present invention discloses a motion vision-based multimodal authentication system and a method thereof. The system comprises several components, including an information-capturing module, an AI module, a memory unit, a processor unit, and a temporal authentication module. The system captures information data from the user and the user's surroundings, categorize the data into identity and predetermined categories, generates login credentials or authorization keys, and matches the real-time information data for user authorization.

This invention provides a system and method for the automatic assembly of screws on large antenna arrays. The system includes an assembly mechanism, which includes an end effector; a linear motion mechanism, on which the assembly mechanism is mounted and moves along a first horizontal direction under the drive of the linear motion mechanism; a lifting mechanism, disposed on one side of the linear motion mechanism along a second horizontal direction; the antenna array can move vertically under the drive of the lifting mechanism; a vision mechanism, fixed to the end effector, for acquiring feature images of screw holes on the antenna array; and a control unit, electrically connected to the linear motion mechanism, the lifting mechanism, the assembly mechanism, and the vision mechanism. The control unit controls the linear motion mechanism and the lifting mechanism to adjust the relative position of the end effector and the screw holes on the large antenna array, and completes the precise alignment of the end effector and the screw holes based on the feature images of the screw holes acquired by the vision mechanism, so that the end effector completes the screw tightening assembly.

The invention discloses an unmanned aerial vehicle countering autonomous tracking system based on visual detection, relates to the technical field of unmanned aerial vehicle countering, and constructs a unique rotor visual fingerprint of each unmanned aerial vehicle by capturing rotor micro-motion visual features such as rotation frequency, blade swing amplitude, rotor blade reflective features and blade edge micro-deformation of the rotor of the unmanned aerial vehicle. The feature processing module performs normalization and dimensionality reduction processing on extracted features, effectively unifies different feature dimensions and eliminates redundant information, and the autonomous tracking module uses rotor visual fingerprints as core matching features and adopts a cosine similarity algorithm and a linear pre-judgment formula to realize accurate target identification. Precise identification and continuous and stable tracking of an unmanned aerial vehicle target are realized, the dual verification module combines basic features of an unmanned aerial vehicle body, and the reliability of a tracking result is further enhanced by presetting a matching threshold value and calculating a comprehensive matching degree through weighting.

This invention relates to the field of brain-computer interface (BCI) data processing technology, and discloses a BCI data processing method that integrates steady-state motor visual evoked potentials and motor imagery. The method includes: generating a bimodal task configuration table; generating a bimodal synchronized stimulation sequence; acquiring and labeling multi-lead EEG signals; extracting visual response features and motor imagery response features; performing motor intention recognition; and generating rehabilitation feedback information. Compared to the single visual evoked paradigm in existing technologies, especially under conditions of significant differences in subjects' motor imagery abilities and low EEG noise ratios, this invention addresses the technical problem of failing to achieve stable motor intention recognition. By synchronously binding motor visual stimulation with motor imagery tasks and jointly extracting visual frequency response and motor imagery desynchronization features, stable recognition of motor intention is achieved, improving the recognition accuracy of rehabilitation BCI training.

This invention discloses an anesthesia injection device, its automatic control method, and a hair transplant robot, relating to the field of medical surgical equipment technology. The device includes: a fixing mechanism for mounting on a robotic arm and being driven by the robotic arm; a first feeding mechanism disposed on the fixing mechanism; an injection mechanism disposed on the first feeding mechanism, the first feeding mechanism driving the injection mechanism to move; and a vision control unit disposed on the fixing mechanism for acquiring images and controlling the movement of the first feeding mechanism and / or the robotic arm based on the acquired images, so that the injection mechanism reaches a target position. The anesthesia injection device, its automatic control method, and the hair transplant robot of this invention can automatically inject anesthetic into the scalp of a target person with low pain during injection, and the injection process is fast and efficient.

The invention provides an optimization method and system for an intelligent data acquisition process, and the method comprises the steps: executing a data acquisition process: generating a data annotation task based on a preset data acquisition plan and a user request sent by a remote client, and carrying out the matching of idle mechanical arms in a mechanical arm cluster; based on the action stream data sent by the remote client, the mechanical arm is controlled to execute the corresponding action, and the motion visual information of the mechanical arm is collected; generating video stream data based on the motion visual information of the mechanical arm, and ending a data acquisition process; performing quality auditing on the video stream data based on a preset quality evaluation strategy to obtain a quality auditing result; performing statistical analysis on the data acquisition process based on the quality audit result to obtain a data acquisition process energy efficiency monitoring report; and optimizing the data acquisition process based on the data acquisition process energy efficiency monitoring report. According to the method, the data acquisition quality is improved by performing quality auditing on the data acquisition result and combining with an optimized feedback mechanism.

This invention discloses a system and method for reviewing and analyzing motion visual data based on ball-hitting triggers, relating to the field of motion multimodal data synchronization and intelligent analysis technology. The system includes a racket trigger module, a head-mounted visual acquisition module, and a data processing terminal, which work together to form an inseparable core architecture of "ball-hitting trigger - data acquisition - synchronized review and analysis." Specifically, the head-mounted visual acquisition module acquires the athlete's first-person perspective scene in real time, annotates it with a high-precision timestamp, and streams it to the data processing terminal in real time, without setting up a large local cache to reduce device load; the racket trigger module acquires ball-hitting related inertial navigation data, completing ball-hitting detection and timestamp generation and uploading through two independent and non-conflicting methods; the data processing terminal caches the received data, triggers deduplication and time-series alignment, and the time-series alignment adopts the nearest original frame matching strategy to preserve the authenticity of the original data and avoid data distortion. This invention enables high-precision synchronization of hitting action and visual data, supports both automated analysis by the host computer and manual review / autonomous analysis by the user, and features rapid data transfer and stable storage capabilities. It solves the technical problems of multi-source data timing misalignment, chaotic trigger signals, high load on head-mounted devices, and poor adaptability in existing technologies. It can be flexibly adapted to various racket sports such as table tennis, badminton, and tennis, providing data support for professional training in racket sports, while reserving space for parameter adjustment in subsequent research and development.

An inertial measurement unit (IMU) is fixed in a known transformation relative to a vision system sensor. The IMU in vision system sensor are mounted to an agricultural machine. The agricultural machine is controlled to move and the orientation of the vision system sensor, relative to a reference point on the agricultural machine, is generated based upon an IMU signal generated by the IMU during the motion. The orientation is then used to control the agricultural machine.

The invention discloses a decoking device of a garbage incinerator, which comprises a lifting frame assembly used for positioning and self-locking the decoking device; the self-adaptive positioning device is used for driving the lifting frame assembly to do retraction movement; the visual module is used for acquiring three-dimensional information of a decoking working position and generating a most suitable tail end posture for decoking; the speed reduction rotating assembly is used for driving the rotating positioning frame to rotate and position; the large-torque driving joint assembly is used for driving the moving telescopic arm assembly to swing up and down and be positioned; the motion telescopic arm assembly is used for adjusting and positioning the spatial pose of the dual-motor vibration tail end assembly; and the dual-motor vibration tail end assembly realizes rapid decoking through high-frequency vibration. The device can be quickly positioned on any section of a hearth, online decoking without shutdown of the incinerator or quick decoking in short-time shutdown of the incinerator is realized, so that the shutdown time is remarkably shortened, the power generation loss is reduced, personal injury is avoided, and long-period, high-load and low-emission stable operation of the incinerator is realized.

The application discloses a quartz tube surface protection type chain plate conveying device and relates to the technical field of quartz tube conveyors. The device comprises a support, a rotating chain plate conveying belt installed on the support, a fixed frame installed on each chain plate of the chain plate conveying belt, a quartz tube body arranged in the fixed frame, a sliding sleeve penetrating through both ends of the fixed frame, a sliding shaft slidably connected to the sliding sleeve, a first air bag and a second air bag arranged at both ends of the sliding shaft, a rotating cylinder rotatably arranged on the support, three electric push rods rotatably connected to the rotating cylinder, a buckle plate rotatably connected to the extension end of the electric push rod, and a visual camera arranged on the buckle plate. The second air bag is inflated to limit the quartz tube body and avoid collision. The electric push rod is elongated, so that the buckle plate moves with the fixed frame, the sliding block drives the visual camera to move, the visual camera comprehensively scans and records the rotating quartz tube body, and the three buckle plates work in turns, so that the working efficiency is improved.

The invention belongs to the technical field of computer vision and unmanned aerial vehicle application, and discloses a self-adaptive key frame extraction method and system based on multi-sensor fusion and motion trail prediction. Synchronously acquiring an image sequence and multi-source sensor data of the unmanned aerial vehicle; calculating motion state parameters of the current frame and the previous key frame, and predicting a future trajectory of the unmanned aerial vehicle; extracting visual features of the current frame and calculating information entropy; dynamically adjusting weights of motion, vision and trajectory prediction factors according to the context of the flight scene through a self-adaptive evaluation model so as to calculate a key frame evaluation score; and finally, determining whether the key frame is extracted based on a dynamic threshold and redundancy verification. According to the method, sensing and prediction information is fused, and a dynamic weight mechanism is introduced, so that the defects of inaccurate key frame extraction and high redundancy in a complex flight state and a changeable scene in the prior art are overcome, and the three-dimensional reconstruction efficiency and the model quality of the unmanned aerial vehicle can be remarkably improved.

The invention discloses an intelligent sports timing running system based on fusion of an improved motion vision technology. According to the system, an image acquisition module, an image preprocessing and enhancing module, a dynamic face detection and attitude robustness module, a dual identity verification module, a dense scene multi-target detection module, a multi-target tracking and data statistics module and a data storage and interaction module are called; the existing technical problems generally existing in a campus timing running system are solved. Aiming at the problem of insufficient face recognition precision in a motion state, the feature extraction capability of the model on dynamic, non-standard postures and low-definition images is improved, and the dependence on a single face feature is reduced in combination with a multi-modal technology. Meanwhile, aiming at the problem of multi-target recognition failure of a dense scene, feature conflicts caused when multiple targets shield and enter a recognition area at the same time are solved, the hardware coverage range is expanded, the multi-target processing capacity of an algorithm is improved, and meanwhile, people stream pressure is dispersed through a scenarized strategy.

The application belongs to the technical field of computer vision and unmanned aerial vehicle application, and discloses a kind of adaptive key frame extraction method and system based on multi-sensor fusion and motion trajectory prediction.Synchronously acquire the image sequence and multi-source sensor data of unmanned aerial vehicle;Calculate the motion state parameters of the current frame and the last key frame, and predict the future trajectory of the unmanned aerial vehicle;Extract the visual features of the current frame and calculate the information entropy;Through an adaptive evaluation model, dynamically adjust the weights of motion, vision and trajectory prediction factors according to the flight scene context to calculate the key frame evaluation score;Finally, based on the dynamic threshold and redundancy verification, decide whether to extract as a key frame.The application fuses perception and prediction information, and introduces a dynamic weight mechanism, overcoming the defects of the prior art in extracting key frames under complex flight conditions and variable scenes, such as inaccuracy and high redundancy, and can significantly improve the efficiency and model quality of unmanned aerial vehicle three-dimensional reconstruction.

This invention discloses a machine vision-based judgment method for blade tip identification in airspace radar, belonging to the field of wind turbine safety monitoring technology. It includes: real-time acquisition of radar ranging data synchronized with image sequences; online quantitative evaluation of the data quality of the radar ranging data to generate a condition assessment result; and adaptive switching between standard and auxiliary working modes based on the assessment result. In the auxiliary working mode, the image sequence is processed to extract real-time motion visual features, which are then compared with a pre-constructed reference visual model to perform point-by-point reliability verification of the radar ranging data, ultimately generating a filtered, highly reliable airspace measurement value. This invention can effectively distinguish between genuine blade signals and interference pseudo-signals, significantly reducing the false alarm rate and improving the reliability of the airspace monitoring system in complex environments.

The application relates to the technical field of sports training methods, and discloses a sports visual training method and system based on a naked-eye 3D display, which comprises the following steps: determining a training item according to a training target; reading preset scene construction parameters, depth-of-field control parameters and training action types in the training item; determining a target depth-of-field presentation mode according to the depth-of-field control parameters; matching corresponding peripheral unit according to the training action types, wherein the peripheral unit comprises a feedback collection unit and / or an auxiliary stimulation unit; and calling a generation algorithm to generate a corresponding training scene based on the scene construction parameters and the target depth-of-field presentation mode, dynamically adjusting the training difficulty through an adaptive algorithm, and realizing progressive training intensity improvement. The method can make the trainer comprehensively train through an ordinary naked-eye 3D display device, can comprehensively stimulate various functional modules of a visual system, and can greatly improve the overall training effect through organic combination of various training modules.

The invention relates to an underwater active retracting and releasing device with power, which comprises a top frame, a retracting and releasing cage, a propelling system, a visual system and a control system, the top frame system comprises a main suspension arm, an arched support beam, a main support frame, a sliding suspension ring, a cross rod, a cross rod hanging plate, a buoy and a stabilizer bar; the folding and unfolding cage is of a modular structure and comprises a replaceable U-shaped folding and unfolding cage and a replaceable V-shaped folding and unfolding cage. The propelling system is composed of four propellers and an installation sleeve, and underwater movement of the device is achieved. The visual system comprises a binocular visual camera and a monocular visual camera and is used for environment perception and target identification; the control system is composed of a shore-based control system and a ship-based control system, remote control and autonomous control are achieved, the device can select a common mode or an active wave compensation (AHC) working mode according to the sea condition, laying and recycling operation of AUG and other ocean equipment is efficiently completed, and the device has the advantages of being high in adaptability, convenient and fast to operate, stable in folding and unfolding and the like. And the automation level and the intelligence level of marine equipment folding and unfolding are remarkably improved.

The invention discloses a thunder-vision fusion method considering road information and sensor characteristics in a roadside perspective, and relates to the field of environmental perception, and the method comprises the specific steps: obtaining radar data and visual data; performing static target judgment on the visual data by utilizing a dynamic threshold mechanism to obtain motion visual data; performing lane positioning on the radar data and the motion visual data, and performing target association on the radar data and the motion visual data of the same lane to form associated data; and inputting the associated data into a trained track adaptive fusion model, and outputting a fused target state track. According to the method, a pure vision static target judgment algorithm is introduced, and a long and short term judgment mechanism and a dynamic threshold are combined, so that the inherent defect that a radar is insensitive to a static target is effectively overcome, and misassociation caused by the asymmetric detection capability of a sensor is reduced.

The application relates to the fields of modern industry and visual intelligence, and discloses a high-precision motion visual positioning and compensation method and system based on multi-view synthesis, which comprises the following steps: acquiring multi-view images and carrying out multi-view fusion to obtain spatial perception information; analyzing a line-of-sight occlusion area, predicting a moving path of a target, extracting target features, and determining an occlusion period; if the occlusion period exceeds a preset threshold, extracting adjacent-view supplementary features and target features, carrying out weighted fusion, and obtaining complete multi-view fusion features; calculating the relative displacement of the target in a three-dimensional coordinate system, denoising to obtain a displacement correction value if the physical movement law is met; generating an optimized tracking path according to the correction value for tracking and prediction adjustment; if the path has local deviation, filtering adjustment is carried out to determine the final three-dimensional position coordinates; and judging the stability according to the coordinates, and determining the tracking result if the stability is met. The method can solve the problem of large positioning deviation in a dynamic environment in the prior art.

The invention discloses a newborn abnormal motion pattern recognition method based on deep learning, and relates to the technical field of computer vision, and the method comprises the following steps: synchronously collecting multi-mode motion data of a newborn in a supine state, the multi-modal motion data comprises a visible light video, an infrared thermal imaging video and a pressure distribution signal from a sensor array below the mattress; according to the invention, through collaborative acquisition and registration of multi-modal data, the interference of package shielding and body position change on video analysis is effectively overcome; through fusion of pressure sensing features and enhanced micro-motion visual features, comprehensive and stable characterization of neonatal micro and specific motion is realized. By introducing a development stage self-adaptive mechanism, the recognition precision and generalization ability of the newborns of different month ages are remarkably improved, and therefore an automatic and high-reliability abnormal movement recognition means fitting the physiological characteristics of the newborns is provided for clinic.