82 results about "Line matching" patented technology

The invention relates to the technical field of medical cosmetology and computer vision, in particular to a 3D skin measurement method based on high-precision measurement of a face concave-convex degree region, and the method comprises the following steps: arranging a structured light camera to collect a face RGB image and depth data, and carrying out the filtering and denoising to generate a smooth point cloud; detecting an RGB image based on FaceMesh to obtain face key points, and constructing a multi-region segmentation algorithm; segmenting an RGB local area and converting the RGB local area into point cloud data; iDW interpolation is carried out to generate a dense point cloud, and a two-dimensional contour line is extracted through a Marking Square algorithm; and mapping the color of the matched color card to an image area to generate a contour map. According to the method, the face recognition model is constructed through the FaceMesh architecture, accurate positioning of multiple face types is adapted, two-dimensional contour mapping is adopted to three-dimensional contour mapping, data are intercepted, local processing is combined, the operation load is reduced by 60%, and efficiency is remarkably improved.

The invention discloses a two-dimensional laser radar positioning method based on intersection point constraint, and belongs to the technical field of robot positioning. The method comprises the following steps: acquiring point cloud data of an indoor environment based on a two-dimensional laser radar; extracting straight line segments from the point cloud data by adopting a seed region growing method; based on a self-adaptive axis selection strategy of a line segment main extension direction, straight line segment category division is carried out, and a straight line matching relation of adjacent frames is established according to intersection point information of the straight line segment and a coordinate axis; and on the basis of the straight line matching relationship of adjacent frames, according to the distance residual error from the minimization point to the straight line of the Gaussian Newton equation, obtaining the pose information of the robot. According to a matching result, a Gaussian Newton method is used to minimize a distance residual error from a point to a straight line to solve a robot pose; the whole method is low in calculation time consumption, can still keep high efficiency under the high cloud density, is suitable for a low-power-consumption processor, and meets the requirements of an indoor robot for light weight and real-time performance.

Implementations claimed and described herein provide systems and methods for map routing matching system, wherein the one or more machine-learning models scores link candidates of location data points to determine a most probable route. In one implementation, use a machine-learning model to calculate a probability score for a plurality of link candidates. The machine-learning model scores each link candidates based on weighting set by training data, wherein the weighting associated with at least one of travel time, match distance, heading difference, speed conformity, road curvature, road classification, or travel distance versus time delay. A plurality of total probability scores derived from one or more combinations of probability scores for respective link candidates is calculated, wherein one link candidate is selected for each location data point.

The invention relates to a laser radar point-to-line weighted matching method based on edge point quality evaluation, and belongs to the technical field of measurement. The method solves the problems that in traditional point-to-line matching, due to the fact that the quality of edge points is not evaluated, the pose estimation precision is low, and the robustness of an unstructured scene is poor. According to the technical scheme, a preprocessing module screens candidate edge points through projection point cloud, target segmentation and roughness calculation; the position judgment module classifies the candidate points as middle, left or right edge points according to the category consistency of the adjacent points; the direction consistency judgment module calculates a direction standard deviation and screens effective edge points; and the dynamic weighting module allocates weights according to an exponential attenuation function, and optimizes the pose through weighting residual errors. The method has the advantages that the pose estimation precision is remarkably improved, the weak structure environment adaptability is enhanced, the calculation efficiency is optimized, and redundant iteration is avoided.

The invention provides a method and a system for accurately positioning a first beam section of a segmental beam by a short line matching method, which are used for optimizing a pier-side bracket opposite-pulling deformed steel bar tensioning construction process, enhancing the stability of a bracket and improving the mounting precision and the construction efficiency of the first beam section; before the first beam section is adjusted, the bracket is pre-pressed, and the influence of non-elastic deformation of the bracket on the positioning accuracy of the first beam section is eliminated; by controlling the axis deviation and flatness of the middle supporting legs, the influence of unbalance loading generated by the kiloton bridge girder erection machine on the line shape of the first beam section is avoided; the centering rod of the prism is regularly self-checked and corrected, so that measurement errors caused by inclination of the centering rod are eliminated; elevation control encryption adopts a zenith distance method, so that the precision is high, and linear control of segmental beam erection is facilitated; before a front auxiliary supporting leg of the bridge girder erection machine is in place and after a pier top block is hoisted, multi-measurement-back displacement monitoring is carried out on a pier body, and the accuracy of a line shape under the working condition is guaranteed. According to the method, the influence of each factor on the positioning precision of the first beam section is eliminated, so that the positioning precision of the first beam section is controlled within a standard allowable range.

The invention relates to an AdobeAudition (AU) extension program-based automatic track matching system and method, and relates to the technical field of audio processing, a picture book analysis module identifies and analyzes a role line sequence and an interval parameter of a line book through a regular expression and a role tag; the audio matching and loading module is used for matching role audios according to a filename rule or a metadata label and loading the role audios to an initial track; the silence detection and marking module adds a mark at a non-silence initial position by adopting framing energy calculation; the line-audio verification module is used for realizing audio and line matching degree analysis through voice recognition to text and font pinyin double-path similarity calculation, and providing a visual calibration interface; and the automatic execution module supports a full-automatic sequential splicing mode or a semi-active sentence-by-sentence splicing mode. The method comprises the steps of line book analysis, audio matching and loading, mute detection marking, verification and calibration and dual-mode execution. The problems of low manual rail alignment efficiency and poor precision in the audio book making process are solved.

The invention discloses a terminal strip drawing terminal identification and cable connecting line matching method and system, and the method comprises the steps: obtaining the layer, geometric shape, text and other information of a terminal strip CAD graph, and extracting the needed line segment and text; training a target detection model, identifying a terminal strip area in the terminal strip drawing, and cutting out a sub-graph; adopting a sliding window method to cut the terminal row into a plurality of sub-graphs with a certain overlapping area; labeling table cells in the sub-graphs, making a data set, and training an RT-DETR model; inputting the sub-graph into the trained RT-DETR model for prediction, outputting cell coordinate information in the sub-graph, and restoring the cell coordinate information to the original graph; all cable connecting lines connected with the terminal strip area are calculated, and the cells representing the terminals are matched with the cable connecting lines according to the geometrical relationship. According to the invention, the matching relation between the terminal and the cable connecting line can be effectively detected, the structured data can be automatically generated, and rapid extraction and analysis of drawing information can be realized.

A computer that includes a processor and a memory, the memory including instructions executable by the processor to determine top key points from one of an aerial feature map or one or more ground feature maps and project the top key points as corresponding lines on the other of the aerial feature map or the one or more ground feature maps. The memory includes instructions to determine a depth estimate for the top key points on the corresponding lines. A high-definition estimated three degree-of-freedom pose of a ground view camera is determined in global coordinates by iteratively determining geometric correspondence between the top key points and the corresponding lines until a global loss function is less than a user determined threshold.

A processor, computer system, and method for hierarchical cache flush. The processor includes a first core and a last level cache. The first core includes a decoder, a memory order buffer, and a first core internal cache. The decoder, in response to an instruction of an instruction set architecture specifying a key identification code and an address for a hierarchical cache flush, generates at least one microinstruction. According to the at least one microinstruction, a flush request with the specified key identification code and the specified address is supplied through the memory order buffer to the first core internal cache, and then from the first core internal cache to the last level cache, so that the last level cache flushes a matching cache line matching the specified key identification code and the specified address.

The invention discloses a flexible automatic assembly line system adaptive to multiple specifications, and relates to the technical field of flexible automatic assembly lines, the flexible automatic assembly line system comprises a main conveying belt and an auxiliary frame, the two sides of the main conveying belt are each provided with a connecting frame, the top of each connecting frame is provided with a top plate, and the top plate is internally provided with a visual detection module used for intelligent identification. According to the flexible automatic assembly line system adaptive to multiple specifications, a scanning result can be transmitted to the information acquisition module, acquisition and local storage can be performed by using the information acquisition module, information backup processing is realized, and then information comparison is performed through the recorded information comparison module; and then the corresponding assembly line module can be obtained by utilizing the assembly line matching module through comparison, the obtained scheme information is finally transmitted to the control module, automatic control can be realized by utilizing the design of a controller in the control module, and different assembly lines can be customized according to the requirements of the intelligently identified flexible assembly parts with different specifications.

The invention provides an autonomous vehicle sharing simulation method and device under a V2G background, and relates to the technical field of man-vehicle matching, and the method comprises the steps: obtaining urban road network data and travel data; performing model construction under a V2G background through the urban road network data and the travel data, and obtaining an automatic driving vehicle sharing simulation model based on interaction of demand response type travel service and an improved O2DES simulation framework; real-time path planning is carried out through a real-time path planning algorithm and the self-driving automobile sharing simulation model, and a path planning result is obtained based on parallel computing, bidirectional searching and path tracking; and carrying out man-vehicle online matching based on the path planning result to obtain a matching result of the autonomous vehicle and the customer. According to the method, the problems that the influence of various dynamic random interactions and the V2G charging and discharging background are not considered in the existing automatic driving electric vehicle sharing simulation and the overall efficiency of the existing simulation is low are solved.

The embodiment of the application provides a shield tunneling machine cutter head water jet cleaning test method and a shield tunneling machine, wherein, in the method, the Hungarian distribution algorithm performs line comparison, cross-line matching and minimum cost relationship establishment on a time matrix, so that a clear minimum time correlation is formed between nozzles and sub-areas, and a calculable balanced relationship of cleaning load is formed among multiple nozzles; the particle swarm optimization algorithm is used to perform fine adjustment of nozzle mounting points, so that nozzle coordinates can be continuously updated within a search range after displacement in a fixed direction; the position of a group of particles is iterated to find a higher reachability of a jet posture under distance and angle conditions, so that the spatial relationship between nozzle pointing, action area normal vector and cutter head structure is updated in multiple rounds, higher coverage matching accuracy is obtained, the nozzle start time and the time amount of the assigned sub-area are formed into a unified time axis and are aligned with a rotation reference, the nozzle start-stop behavior and the cutter head position form a continuous and recordable cleaning structure.

A method for offline perceptual motion inference is provided that may include obtaining map data indicative of an environment and obtaining data associated with at least one agent. The method may include determining a trajectory for the agent and matching the trajectory of the agent to a lane connection line. The method may also include determining traffic light parameters. A system and a computer program product are also provided.

The application discloses a power pole inclination detection and early warning method and related equipment, first, the power pole image is collected by a shooting device and corrected, and then the corrected image is input into a power pole image segmentation model composed of a multi-layer channel feature extraction network MCFNet, a hollow convolution bar pooling module DCSP and a channel attention module CE. MCFNet extracts semantic information from channels layer by layer, reduces calculation parameters and improves efficiency; DCSP uses hollow convolution and bar pooling to expand the receptive field and improve the shape feature recognition effect of the power pole; CE suppresses redundant channel information and highlights effective features. The model can cope with complex environments and overcome the low precision of existing models. Then, the polar line matching algorithm is used to calculate the left and right side edge disparity of the power pole, a three-dimensional point cloud is constructed, and the inclination angle calculation precision is improved. Once the inclination angle exceeds the preset threshold, a warning is sent within a few seconds, the safe operation of the power transmission tower is ensured, and the accident probability is reduced.

Implementations claimed and described herein provide systems and methods for map routing matching system, wherein the one or more machine-learning models scores link candidates of location data points to determine a most probable route. In one implementation, use a machine-learning model to calculate a probability score for a plurality of link candidates. The machine-learning model scores each link candidates based on weighting set by training data, wherein the weighting associated with at least one of travel time, match distance, heading difference, speed conformity, road curvature, road classification, or travel distance versus time delay. A plurality of total probability scores derived from one or more combinations of probability scores for respective link candidates is calculated, wherein one link candidate is selected for each location data point.

According to embodiments of the present disclosure, methods, apparatuses, devices and media for lane line fusion are provided. The method includes obtaining first lane line information associated with a first sensor and second lane line information associated with a second sensor; determining first matching information between a first set of lane lines indicated by the first lane line information and a set of reference lane lines indicated by reference lane line information; determining second matching information between a second set of lane lines indicated by the second lane line information and the set of reference lane lines indicated by the reference lane line information; and fusing the first lane line information and the second lane line information based on the first matching information and the second matching information. Thereby, the success rate of lane line matching can be improved in special road sections and lane line mis-detection, and the accuracy and robustness of lane line matching in all scenarios can be improved.

The invention relates to a four-layer monolithic three-dimensional integrated computing device and a control method thereof.The four-layer monolithic three-dimensional integrated computing device sequentially comprises a logic control layer, a resistive memory computing layer, a near memory data processing layer and a near-end cache preprocessing layer from bottom to top, and the logic control layer is responsible for generating an access control signal and a data scheduling strategy and driving the resistive memory computing layer to operate; the resistive random access memory computing layer executes dictionary table lookup based on the received table lookup task, or obtains a multi-line matching result through a parallel comparison strategy in a preset matching scene; the near storage data processing layer stipulates and combines multiple rows of matching results, and outputs stipulated matching results; the near-end cache preprocessing layer stores the protocol result, and completes data rearrangement, packaging and preprocessing of the to-be-processed data according to the received scheduling instruction, so that the problems of high dictionary table lookup delay, high energy consumption, relatively low vector retrieval bandwidth, slow response and the like in related technologies are solved; delay and energy consumption of dictionary table lookup are reduced, and vector retrieval bandwidth, response speed and energy efficiency ratio are improved.

The present invention relates to a method for self-vehicle positioning using lane line matching, which includes the steps of: obtaining a camera fitting curve; the camera fitting curve is in the camera coordinate system; extracting camera shape points; packing to obtain a set of camera shape points; obtaining map lane lines; packing to obtain a set of map shape points; performing secondary processing on the set of camera shape points; performing secondary processing on the set of map shape points to convert the map lane lines into map lane line fitting curves; performing equally spaced shape point sampling to obtain a set of sampled camera shape points and a set of sampled map shape points; calculating the mean square error to obtain the mean square error of the sampled camera shape points and the mean square error of the sampled map shape points; performing ICP matching to obtain R and T error parameters; obtaining the corrected vehicle position information. The present invention achieves the technical effects of low cost but high precision; it can handle almost all road conditions with extremely small errors.

The application discloses a visual positioning method and device, a storage medium and an electronic device, and relates to the technical field of information. The method comprises the following steps: acquiring a current frame image and a previous frame image corresponding to the current frame image; extracting feature points and feature lines corresponding to the previous frame image and feature points and feature lines corresponding to the current frame image, and performing feature point matching and feature line matching on the previous frame image and the current frame image to obtain a feature point matching pair and a feature line matching pair; constructing a point reprojection error and a line reprojection error between the previous frame image and the current frame image based on the feature point matching pair and the feature line matching pair; constructing a target optimization function according to the point reprojection error and the line reprojection error, and solving a relative pose between the current frame image and the previous frame image; and performing visual positioning on a target object according to the relative pose. The application can overcome the limitation of feature point extraction in a weak texture environment, thereby improving the accuracy and robustness of visual positioning.

The invention provides an interactive motion capture and personalized work generation method for virtual reality, and belongs to the technical field of interactive capture and generation, and the method comprises the steps: carrying out the mining of user demands, and obtaining a demand line set, matching a coarse scene matched with each demand line and a fine scene matched with each demand anchor point in each demand line from a scene building database, and carrying out scene building on all the coarse scenes and the fine scenes to obtain a virtual reality scene; in the virtual reality scene, a response interaction matrix of the user responding to each preset output is captured, the response interaction matrix and a preset interaction matrix are analyzed, meanwhile, a self-defined interaction matrix of the user in a non-preset output stage is captured, and the self-defined interaction matrix and an analysis result are subjected to complementary processing to obtain key design features of different demand lines; and carrying out supplementary processing on the key design features to generate personalized works and outputting the personalized works. And the core pain point of disjunction of art creativity and technology realization in digital media art VR creation is solved integrally.

The application relates to a space non-cooperative target position and posture measurement method based on binocular event flow. The method comprises the following steps: building a binocular event camera according to the left-right fixed connection installation of two event cameras; photographing a target in motion by using the binocular event camera; projecting the left and right event flows collected in a fixed time interval into a three-dimensional space; reconstructing a target three-dimensional straight line according to a straight line; obtaining a projection straight line according to a pre-set projection relationship; matching events in the left and right event flows with the projection straight line according to the nearest neighbor principle; calculating the distance from the left and right event flows of the binocular event camera to the corresponding projection straight line according to the event-straight line matching result; and solving the space non-cooperative target position and posture by minimizing the distance through a nonlinear optimization algorithm. The method can adapt to the space non-cooperative target position and posture measurement in a complex space environment.

The present invention relates to a vision feature matching and positioning method for driverless container trucks in ports. The method includes the following steps: S1: The vehicle travels on a preset road, acquires a current road condition picture based on an in-vehicle camera, performs image thresholding processing on the current road condition picture, and outputs target picture data information; S2: Based on the target picture data information, deep learning is carried out, feature transformation is performed on the semantic segmentation data matrix, and the relative pose of the vehicle is output; S3: Coupling the high-precision map data information, respectively according to the lane line matching and positioning algorithm and the Berth No. matching and positioning algorithm, the first vehicle pose and the second vehicle pose are output; S4: Coupling the relative pose of the vehicle, the first vehicle pose and the second vehicle pose, the corrected vehicle pose and confidence information are output. The present invention can achieve stable positioning of driverless container trucks in special scenarios such as yards and quay cranes in the port environment, providing a reliable guarantee for the stable operation of driverless container trucks in the port.

The invention discloses a machine vision board line distinguishing and edge banding machine ink-jet printing system. A visual collection unit collects board lines, a data processing unit extracts multi-dimensional texture features and generates dynamic edge banding patterns and two-dimensional codes, an edge banding execution unit carries out edge banding through a white edge banding belt, and an ink-jet printing unit carries out printing after scanning the codes. The system solves the problems that the edge band lines are not unified with the plates and personalized customization is limited, achieves accurate line matching, reduces edge band reserve and is suitable for processing of high-grade plate furniture.

A line feature matching method based on dynamic matching and absolute median difference includes the following steps: 1) receiving RGB frames captured by an RGB-D camera as visual input images; 2) extracting line features using the EDLines line detector; 3) parameterizing line features using endpoints; 4) describing the parameterized line features based on discrete key points; 5) dynamically matching line features extracted from previous and next frames using a dynamic programming algorithm; and 6) estimating the standard deviation of displacement changes for all line feature matches using the absolute median difference method, and using this estimate to eliminate erroneous or low-quality line matches. This method can extract a sufficient number of correct line feature matches in scenes with occlusion, rotation, and appearance changes.

The invention discloses a digital map and infrared image accurate registration method based on contour line matching, which comprises the following steps of: 1, acquiring an infrared image in front of an airplane as an image to be registered, and taking a terrain simulation image of an airborne digital map as a reference image; 2, mapping the infrared image into a terrain simulation image, and carrying out LOD refinement on a mapping area of the infrared image to obtain the terrain simulation image fused with a texture mapping; 3, respectively extracting contour lines L1 and L2 from the infrared image and the terrain simulation image fused with the texture map by using a contour line extraction method; and 4, adjusting data points on the image by using external parameters, and then carrying out precise registration on L1 and L2. According to the method provided by the invention, the detection problem of a given curve in an original image can be converted into the problem of searching a peak value in a parameter space, the accurate registration of a virtual image and a real shot image is realized, and the environment perception capability of a pilot is improved.

The application provides a vortex reconstruction and tracking method, aiming at the defects of the traditional vortex reconstruction method in three-dimensional sampling, pseudo matching suppression and the like, the application effectively matches vortex axes in three-dimensional directions, takes into account the three-dimensional sampling effect, and has a better reconstruction effect on isotropic vortices; meanwhile, by using the constraint characteristics of the vector direction of Liutex, physical constraint conditions during matching are set, the phenomenon of pseudo vortex axis point matching is effectively reduced; by using a graph neural network and reinforcement learning, vortex axis matching is optimized, three-dimensional reconstruction precision is improved, and pseudo matching is suppressed by global topological constraints; by extracting vortex axis points with dominant absolute rotation intensity, changes in the structure of a single vortex when evolving over time in a flow field are intuitively observed.

The present invention relates to a method for identifying power cables, including steps such as transmitting signals, receiving signals, phase calculation, and line matching determination. The power cable identification method provided by the present invention is based on the double modulation wave technology, utilizes the principle of electromagnetic induction, and realizes the on-line detection of power cables by differentiating different phases of the coupled signals. The transmitting end couples the identification signal to the metal armor of the cable through electromagnetic induction; while the receiving end detects the signal on the metal armor and uses the phase identification technology for differentiation, so as to determine the corresponding relationship between power cables.

The application discloses a K-nearest neighbor power line identification method using comprehensive features, which comprises the following steps: inputting radar target echo to signal detection to obtain strong scattering points, correlating multiple frames of data to obtain a pitch-distance two-dimensional graph and an X-Y two-dimensional graph, identifying a power tower to obtain a power tower position, obtaining power line straight lines of a tower topology, detecting the power line straight lines, extracting power line features, obtaining power line straight line feature vector samples and noise straight line feature vector samples, identifying unknown straight line feature vectors according to an expert knowledge base, and obtaining a power line identification result. The application is based on power line Bragg scattering echo peaks, combined with closely related power tower target identification, extracts comprehensive features, performs power line matching on the power line straight lines obtained through straight line detection and the power line straight lines obtained through the tower topology, and realizes accurate identification of the power line based on a K-nearest neighbor algorithm, so that the correct identification probability is high, the target identification range is wide, and the self-adaptation capability is strong.

The application relates to the technical field of automatic driving, and discloses a lane line matching method and device, which comprises the following steps: identifying lane lines in a video picture, adopting a polynomial curve to represent the lane lines to be matched, and obtaining lane line equations; predicting the similarity between any two lane lines based on the lane line equations; constructing a matching cost function according to the similarity; obtaining a matching cost matrix of the lane lines to be matched through the matching cost function; solving the matching cost matrix to obtain two optimally matched lane lines; and judging whether the two optimally matched lane lines satisfy preset conditions, and matching the two lane lines satisfying the preset conditions as the same lane line at different time points or in different spaces. The application has small calculation amount, faster matching speed, can be used for matching the observation of the same lane line at different time points and in different spaces, has a larger application range, and achieves a simpler effect.