349 results about "Optical flow" patented technology

Optical flow information is determined and used to identify a video clip from among a sequence of frames. The video clip may be identified based on motion features derived in part from the optical flow information. In some embodiments, semantic information is concatenated with the motion features derived from the optical flow information.

This invention relates to the field of control methods for public transportation equipment, specifically to a control method for LED wall washer lights for driver fatigue wake-up in long tunnels. The method includes: calibrating driver fatigue-sensitive sections in each spatial unit of the entire tunnel and completing alignment; updating the vehicle's spatial unit and collected parameters in real time to classify the driver's real-time fatigue state; generating a fatigue wake-up light signal encoding parameter set for the vehicle corresponding to the current driving tracking ID, the current fatigue state, and the current spatial unit; and driving the wall washer lights to generate optical flow signals according to the set parameters. This invention constructs an optical flow signal framework deeply bound to the driving state, simultaneously achieving both fatigue wake-up and directional navigation functions. It also binds the driver's real-time fatigue state classification to a stepped stimulation level, ensuring that the optical flow signal only acts on the surrounding visual area corresponding to the tunnel sidewall, without interfering with the driver's observation. This spatial distribution eliminates the interference of dynamic lighting on normal driving observation.

The application provides a liquid crystal optical flow control chip, an optical imaging device and application, and the liquid crystal optical flow control chip comprises a substrate, the substrate is provided with a closed microstructure, the microstructure comprises a plurality of sample pools and a plurality of optical pools which are the same in number and one-to-one corresponding, each sample pool and one optical pool corresponding to the sample pool are communicated through a micro mixing channel; each sample pool is provided with an inlet and outlet pipeline one, and each optical pool is provided with an inlet and outlet pipeline two; the micro mixing channel comprises a first channel and a second channel which are communicated with each other, wherein the first channel is provided with a sample inlet pipeline. The application is based on the aptamer conformation change principle and the liquid crystal birefringence performance, and further develops the liquid crystal optical flow control chip and the optical imaging device, different liquid crystal optical appearances of samples on the liquid crystal optical flow control chip are observed through a polarizing microscope, and the purpose of detecting exosomes in the samples is realized, and the application has the advantages of high throughput, rapid detection and high sensitivity.

A system and method for distributed temperature and strain sensing using Chirp-Pulse Coherent Optical Time-Domain Reflectometry (CP-COTDR) utilizes optical flow processing. The method involves constructing a two-dimensional (2-D) waterfall data array from collected CP-COTDR frames. Optical flow processing is applied directly to the 2-D data to estimate the local shift of features caused by external perturbations. The optical flow is preferably calculated using a weighted least squares method with an adaptive analysis window based on the local flatness of signal derivatives. This approach provides versatile and high spatial resolution, reduces accumulative errors, and enables efficient data processing compared to conventional cross-correlation techniques.

The present application relates to the technical field of image understanding, in particular to a page element semantic analysis method based on visual language enhancement; a dense optical flow field and inter-frame state transition rule joint analysis is performed on an acquired visual image frame sequence, a two-dimensional spatial region in which a local rendering state is abnormal is identified, and a space-time feature matrix is constructed; a space-time feature matrix and an edge pixel brightness attenuation gradient and a visual darkening feature matrix of a neighborhood environment are extracted, and a pseudo-depth blocking parameter is identified; foreground multi-modal features inside the space-time feature matrix and background multi-modal features of an external environment are extracted, and a relative entropy divergence in a multi-modal semantic space is calculated; a multi-dimensional joint determination model is constructed using the pseudo-depth blocking parameter and the relative entropy divergence, and whether the spatial region is an interference pop-up window is identified; when the identification result is an interference pop-up window, warning data containing attribute information of the interference pop-up window is output, and an adaptive locking mechanism of interface analysis state is triggered.

The application discloses a video target recognition method based on multi-model hot switching, and belongs to the technical field of computer vision and video processing. The method comprises an arbitration module, a switching control module and a feature adaptation and buffer module. The arbitration module generates a switching preparation signal by extracting multi-dimensional indexes such as optical flow mean, local variance, target density and scene confidence in real time through a lightweight channel independent of main reasoning. The switching control module performs atomic replacement of a computation graph pointer in a vertical blanking period, and realizes millisecond-level hot switching with zero frame loss by combining an asynchronous pre-copy and a chasing mechanism of a double buffer. The feature adaptation and buffer module solves tensor shape mismatch between heterogeneous models through a pre-compiled adaptation layer. The application also provides optimization schemes such as multi-index nonlinear fusion decision, zero-copy memory management, local slice focus reasoning and edge-cloud hierarchical unloading, significantly reduces switching delay, guarantees continuous recognition of a video stream, and is suitable for edge computing scenes with limited resources.

The application discloses a kind of live content auditing method and system based on AI, belong to network live content supervision technical field.The method extracts the multi-dimensional characteristics of anchor by cognitive consistency analysis model, calculates the deviation degree of feature correlation matrix to detect cognitive dissonance;Adopt optical flow tracking and LSTM network analysis microexpression infers implicit intent;Monitoring semantic trajectory drift identifies sensitive content;Construct audience group dynamic graph to analyze emotional transmission intensity;Through bayesian network, from audience abnormal reaction, deduce illegal behavior;Evaluate cognitive load to identify information overload cover;Calculate node centrality to implement transmission blocking;Fusion multi-source information outputs rule determination.System includes cognitive feature analysis, microexpression identification, semantic trajectory monitoring, emotional transmission analysis, reverse reasoning, cognitive load calculation, transmission blocking and decision output eight modules, work cooperatively through event-driven mechanism.The application breaks through the surface limit of traditional audit, can identify cognitive fraud, psychological manipulation and other hidden violations, realizes active defense.

The embodiment of the application provides a kind of unmanned plane hovering method and related equipment, belong to unmanned plane technical field.In the application scheme, the movement information of unmanned plane can be detected from ground image sequence in high frequency and real time using optical flow algorithm, and the pose of unmanned plane is quickly adjusted according to the movement information of unmanned plane in time, when the disturbance of unmanned plane appears, it is quickly corrected, so that unmanned plane can be stably hovered above ground marker.

The application discloses a method for classifying grasping states based on visual-haptic learning, builds an experimental scene for the method, and performs grasping experiments with a 6-DOF robot arm equipped with parallel grippers. Each end of the parallel grippers is equipped with a visual-haptic sensor to capture the small sliding trend of the object. The optical flow dataset OFB-6 of deformable objects in three states of over-stability, complete stability and instability is constructed. The published dataset is used for benchmark testing. The recognition accuracy of the Transformer model is improved before and after the addition of the optical flow, verifying the effectiveness of the proposed optical flow guided Transformer model learning. The optical flow method of the application can detect these small changes in advance, thereby timely adjusting the grasping strategy. The dynamic changes in the tactile image can be more accurately captured. Through the fusion of visual and tactile information, the model can more comprehensively understand the physical changes in the grasping process, and improve the overall grasping performance.

This application relates to the field of tunnel construction safety monitoring technology, and discloses an intelligent identification method for water seepage risk at the tunnel face in water-rich clay layers. This method controls an active infrared excitation module to project radiation in an oblique incidence manner, uses polarization imaging to acquire sequences and calculates Stokes vectors; calculates the degree of linear polarization based on the Stokes vectors and performs threshold segmentation to generate a water film distribution mask; uses the Stokes vectors to construct a polarization angle field reflecting local normal changes as texture features, solves the optical flow constraint equation to obtain the instantaneous velocity field; performs frequency domain transformation on the velocity field, extracts the proportion of low-frequency creep energy through power spectral density analysis, and determines the soil rheological softening factor; finally, combines the water film distribution, rheological softening factor, and instantaneous velocity field to construct a mudslide risk index. This invention utilizes polarization texture reconstruction to solve the problem of weak texture tracking and eliminates elastic vibration interference through frequency domain energy decoupling, effectively improving the accuracy of mudslide risk identification.

The present application relates to a kind of based on two-way optical flow to realize the method for fast micro-expression recognition processing, comprising the following steps: according to visual system acquisition tester face micro-expression video segment information;Extract the emotional video segment in emotional memory library, and the facial muscle movement situation of micro-expression in emotional video segment is captured by positive and negative two-way optical flow;Extract method extracts key frame in emotional video segment, and the redundant frame in continuous sequence image is eliminated;Call the optical flow information between key frame in optical flow information memory library.The present application also relates to a kind of two-way optical flow to realize fast micro-expression recognition device, processor and storage medium.The method for fast micro-expression recognition processing based on two-way optical flow of the present application, device, processor and its computer readable storage medium, the facial muscle movement situation of micro-expression is captured by positive and negative two-way optical flow, the micro-expression of tester is identified using muscle movement trend, and the micro-expression recognition accuracy is improved.

This invention provides a method for measuring water flow velocity and related products. The method includes: acquiring a standardized image sequence of a target water area; extracting candidate feature points from every two adjacent real-time acquired images in the standardized image sequence using the SIFT algorithm, and filtering them according to preset screening criteria to obtain a set of candidate feature points; performing optical flow field analysis based on every two adjacent real-time acquired images using the RAFT optical flow model to obtain a dense optical flow field; determining a set of candidate optical flow vectors from the dense optical flow field based on the coordinates of each candidate feature point in the candidate feature point set; and determining the water flow velocity of the target water area based on the set of candidate optical flow vectors and a preset pixel-to-actual distance conversion relationship. This invention achieves accurate capture of the continuous motion trajectory of water flow and reliable matching of stable feature points, effectively improving velocity measurement accuracy and anti-interference capability.

This invention relates to the field of agricultural drone spraying technology, specifically to a method for detecting the spraying effect of agricultural drones based on visual monitoring. The invention acquires canopy images during the spraying process of the agricultural drone, then analyzes the airflow resistance parameters of pixels based on texture and optical flow to determine gap regions; analyzes the pixel-canopy scale coefficient to determine the flow capacity parameters of each pixel, thereby obtaining the airflow impedance coefficient of the pixel; then determines the airflow impact center and obtains the cumulative impedance map of airflow propagation; further, based on the distance between pixels in the gap region and the airflow impact center, and the pixel-canopy scale coefficient, obtains the pesticide penetration index. This invention introduces the pixel-canopy scale coefficient to analyze the visual physical scale of canopy gaps and the airflow resistance, thereby assessing the physical flow capacity of canopy gaps and simulating airflow propagation to assess the difficulty of pesticide penetration through the canopy, thus quantitatively evaluating the spraying effect of the agricultural drone.

This invention discloses a multimodal image real-time registration and fusion system and method for cardiac interventional surgery, comprising: a multimodal image acquisition module, an image data transmission module, an intraoperative twin image registration and analysis platform, an algorithm calculation module, a parameter adjustment module, and an image fusion output module. After acquiring image data, the multimodal image acquisition module sends it to the intraoperative twin image registration and analysis platform via the image data transmission module. The platform incorporates a cardiac deformation adaptive optical flow algorithm, a multi-resolution hierarchical mapping algorithm, and a large-deformation cardiac image registration algorithm. The algorithm calculation module performs registration calculations, and the parameter adjustment module optimizes the fusion parameters based on the calculation results. Finally, the image fusion output module generates and outputs the fused image. This invention achieves real-time and accurate multimodal image registration and fusion, providing reliable image support for cardiac interventional surgery and improving surgical accuracy and safety.

This invention discloses a camera gimbal based on a multi-rotor drone, belonging to the field of drone technology. The drone part includes a center plate main body, a center plate upper plate mounted on top of the center plate main body, landing gear mounted below the center plate main body, and an onboard computer and camera mounted on the center plate upper plate. Sensors are mounted between the center plate main body and the center plate upper plate, and an optical flow meter is mounted below the center plate main body. A 4000mAh 4S large battery is mounted between the landing gear and the center plate main body. The camera of the drone is fixed by a dedicated camera gimbal. A pressure sensor is placed in the center of the camera gimbal's groove. For most monocular and binocular camera models on the market, when placed on the gimbal, the pressure sensor senses the weight and controls the fixing clamps on both sides of the gimbal to automatically clamp the camera. The fixing clamps have a certain angle of change in all directions, allowing the camera to adjust the field of view in all directions (up, down, left, and right).

An image encoding / decoding method and apparatus for performing bi-directional optical flow (BDOF) and a method for transmitting a bitstream are provided. An image decoding method according to the disclosure is an image decoding method performed by an image decoding apparatus. The image decoding method can include the steps of deriving prediction samples of a current block based on motion information of the current block, determining whether BDOF is to be applied to the current block, if the BDOF is to be applied to the current block, deriving a gradient for a current sub-block in the current block, deriving a refined motion vector (v x ,v y ) for the current sub-block based on the gradient, deriving a BDOF offset based on the gradient and the refined motion vector, and deriving refined prediction samples of the current block based on the prediction samples of the current block and the BDOF offset.

This invention relates to the field of defect detection technology in image recognition, specifically disclosing a method and system for detecting defects in mobile phone SIM card tray manufacturing based on image recognition. By acquiring multiple frames of time-series images and performing sub-pixel-level spatiotemporal registration, the influence of mechanical vibration and displacement deviation is eliminated. The optical flow field is calculated, and local strain tensors, optical flow vorticity, and motion consistency entropy are extracted to construct a time-varying feature matrix. Frequency domain decomposition and motion amplification are performed on the feature matrix to enhance the signal response of minute defects. Dynamic time-domain features and static image features are fused to establish a deep learning-based defect detection model. Finally, a defect probability cloud map is generated, and the defect type and location are output using a dynamic threshold mechanism. This invention achieves high-precision and high-efficiency automatic detection of minute defects in mobile phone SIM card trays, exhibiting good stability and adaptability.

This invention relates to the field of image processing technology and discloses a multimodal data fusion method for online quality monitoring of plastics. The method includes: simultaneously acquiring a reference optical image of the plastic surface and a temporal infrared thermal distribution image containing the current frame and the previous frame infrared image; calculating a spatial temperature gradient matrix; correcting the geometric affine matrix based on the displacement vector normal component determined by the dense optical flow field, constructing a spatiotemporal compensation mapping matrix to achieve alignment of heterogeneous pixel coordinate systems, and generating a structure-guided tensor; applying morphological opening operations to remove low-frequency background thermal fluctuations from the structure-guided tensor to obtain a pure-state high-frequency thermal gradient matrix; adjusting the diffusion coefficient using the pure-state high-frequency thermal gradient matrix as a constraint, applying anisotropic diffusion filtering to the reference optical image, and extracting defect regions. This invention achieves physical-level decoupling between optical artifacts and real defects, eliminates feature mapping deviations caused by thermal conduction hysteresis, and improves the signal-to-noise ratio of image feature extraction.

The application discloses an event camera-based power inspection image overexposure repair and enhancement method, which performs adaptive slice integration on an event stream driven by a structure information entropy model, realizes cross-modal accurate alignment of event slices and synchronous frames in combination with optical flow motion compensation, and thus obtains a physically reliable event structure priori. On this basis, the priori is used to guide overexposure mask optimization and a texture joint network to repair, and local frequency adaptive enhancement is performed to highlight defect features. In this way, texture distortion and structure artifacts caused by irreversible loss of information in the overexposure area can be effectively overcome, the repaired image realizes structure continuity and color consistency at the overexposure boundary, defect details are restored with high fidelity and visually enhanced, and finally the quality of power grid safe operation and maintenance is ensured.

The present application relates to the technical field of medical data processing, and provides a medical image data processing method based on deep learning, which comprises the following steps: collecting data according to a collection template, extracting pulse timing by adaptive threshold peak detection, calculating instantaneous phase by linear interpolation, comparing quantitative indicators with preset threshold values, judging steady state by combining peak loss rate and mutation detection rules, and verifying by calculating inter-channel phase consistency; splitting the collected data according to concept entities to form a data relationship model, implementing fast global rigidity estimation, applying affine transformation, estimating pixel-level displacement field by using a pyramid dense optical flow network, applying the displacement field to original pixels, and performing time domain fusion by taking optical flow confidence and registration residual as weights; cropping the short-time steady image sequence after registration compensation, taking a mixed network of a convolution front end and a space-time Transformer backbone as a prediction model, and outputting a pixel-level risk heat map, a candidate lesion list and confidence intervals of each output.

This utility model relates to the field of pipeline inspection robot technology, and discloses a pipeline inspection robot, including a waterproof shell. An optical flow meter is installed inside the waterproof shell. A button and an indicator light are located on the upper surface of the waterproof shell. An auxiliary detection component is located on the upper middle part of the upper surface of the waterproof shell. A heat sink is fixedly connected to the upper left side of the waterproof shell, and a charging port is fixedly connected to the upper left side of the waterproof shell. In this utility model, the robot is first started by the button, then the indicator light displays the status. The optical flow meter measures the movement speed and posture. Then, a CMOS camera and a supplementary light are used to photograph the inner wall of the pipeline. The supplementary light provides illumination when there is insufficient light. The CMOS camera's rotation angle can reach approximately 120 degrees, thus achieving omnidirectional visual coverage for the pipeline inspection robot, improving the accuracy and efficiency of pipeline inspection.

This application discloses a visual tracking method, apparatus, device, and medium for industrial robots, relating to the field of industrial robot technology. The method includes: acquiring a continuous image stream and global motion data of the industrial robot's end effector camera; performing optical flow tracing on the current frame and the previous frame to obtain feature point displacement vectors, and calculating mechanical vibration energy indices based on these vectors; adaptively adjusting a smoothing factor based on the vibration indices, filtering the global motion data to obtain a stable image transformation matrix, and performing a geometric transformation on the current frame image to generate a stable image; inputting the stable image into a Transformer architecture detection model, and obtaining the detection result through cross-attention calculation; extracting the target feature vector of the current frame based on the target bounding box, performing dual checks of occlusion gating and consistency gating, executing a corresponding memory update strategy, and outputting the target tracking result. This method can achieve accurate tracking.

Methods for motion-controllable video diffusion include extracting optical flow fields from an input video and computing warped noise by iteratively warping noise between consecutive frames using the optical flow fields. The iteratively warping includes (i) re-Gaussianizing expanded pixel regions by sampling fresh Gaussian noise, and (ii) aggregating contracted pixel regions by merging noise particles and renormalizing variance to preserve spatial Gaussianity. An output video is generated by initializing a diffusion process with the warped noise and iteratively denoising to produce temporally coherent output frames. Various other methods, systems, and computer-readable media are also disclosed.

A computer program product interacts with machine-readable mediums with instructions for automated launch detection and recognition. It captures a video of a specified region using detectors on a vehicle. Raw image frames from the video undergo pre-processing, followed by feeding through anomalous clutter tracker pipeline, exceedance detection pipeline using global clutter suppression pipeline and an optical flow pipeline. Image windows are generated by rejecting clutter using exceedance pixel locations to create image chips of the target and local background that are run through a convolutional neural network, classifying them as launches or background clutter. A Kalman filter is applied to at least one frame detection list to generate a launch detection list and causes the launch to be detected.