24 results about "Rigid transformation" patented technology

This application belongs to the field of computer vision technology and discloses a cross-source point cloud registration method based on adversarial neural networks. The method includes: inputting a multi-frame LiDAR point cloud sequence and a UAV video sequence; segmenting each frame of the LiDAR point cloud into ground points and non-ground points and registering them to obtain a global LiDAR point cloud; performing motion recovery structure reconstruction on the UAV video sequence to generate a dense color point cloud; inputting the global LiDAR point cloud and the SfM dense color point cloud into an MMAlignNet network to construct a shared latent feature space; solving for the rigid transformation matrix from the LiDAR point cloud to the SfM dense color point cloud; and aligning the cross-modal point clouds according to the rigid transformation matrix to obtain a multimodal 3D reconstruction result. This application improves the registration accuracy, completeness, and robustness of large-scale outdoor environment 3D reconstruction by fusing UAV imagery and mobile LiDAR data, combined with cross-modal feature alignment and geometric constraint optimization methods.

The present application relates to a mobile platform trajectory planning and positioning correction method for large-size measurement scenarios, comprising laying out feature marker points, obtaining feature marker point coordinates in the digital model coordinate system and the actual workpiece coordinate system; solving and optimizing the rigid transformation matrix of the digital model coordinate system and the actual workpiece coordinate system; mapping the theoretical trajectory points in the workpiece CAD digital model to the actual workpiece surface and optimizing, realizing the operation trajectory planning of the mobile platform; based on the chain matrix relationship between each coordinate system, real-time solving the pose matrix of the mobile platform in the global coordinate system and calculating the positioning error optimization, realizing the global positioning of the mobile platform; obtaining the actual trajectory point set of the mobile platform under the binocular vision system for dynamic correction. The present application meets the high-precision and automated operation requirements in large-size measurement scenarios, has high-precision positioning and closed-loop control characteristics, and is suitable for the automated measurement tasks of large-size components in the fields of aerospace, energy equipment, etc.

This invention discloses a method for laying out irregularly shaped decorative lines based on 3D laser positioning and BIM. The method includes the following steps: obtaining virtual coordinates of the design curve from the BIM model; converting the virtual coordinates into actual coordinates controlling the laser angle through a rigid transformation model and generating a projection path; pre-treating and coating a photosensitive layer on the construction surface; projecting a laser using a laser projector to excite a visible colored laying out path on the photosensitive layer; and finally marking and solidifying the path. This invention, by combining BIM modeling with a laser projector, transforms the virtual path into an actual laying out path, achieving full automation and high precision in laying out complex irregularly shaped curves, significantly improving the quality and efficiency of decorative construction.

This invention discloses a method for local feature transfer and fusion of a 3D model, belonging to the field of 3D reconstruction technology. The method includes: loading and parsing the original model and the target model; constructing and caching a mesh topology adjacency list to achieve O(1) complexity neighbor lookup; selecting the corresponding boundary control circle and performing centroid normalization; calculating the optimal rotation matrix through singular value decomposition of the covariance matrix, and constructing a rigid transformation matrix by combining an adaptive scaling factor; replacing the vertices of the local region of the transformed target model with those of the original model to generate a fused model; performing multi-level expansion of the splicing region based on the adjacency list, and applying Taubin forward and reverse smoothing iterations to eliminate seams. This invention ensures registration accuracy and global optimality through SVD decomposition, effectively prevents model volume shrinkage through Taubin bidirectional smoothing, and significantly improves computational efficiency through the adjacency list caching mechanism, achieving high-quality, high-efficiency, and automated local feature transfer and fusion of a 3D model.

This invention discloses a self-supervised method and system for 3D pedestrian reconstruction based on SMPL pose constraints and a hybrid deformation field. Addressing the shortcomings of existing methods, such as the lack of prior knowledge of human structure, difficulties in adapting skin weights, and the inability to balance physical constraints with non-rigid details, this invention uses a normalized Gaussian field obtained through self-supervised optimization as input and proposes three key technologies: 1. A learnable skin weight field that directly predicts the complete skin weight vector of Gaussian elements; 2. A hybrid deformation field framework that concatenates a linear hybrid skin rigid transformation with a six-plane non-rigid residual transformation to achieve synergy between physical plausibility and detailed modeling; 3. Multi-level regularization constraints that ensure the physical plausibility of the reconstruction from three spaces: parameters, geometry, and image. Experiments show that this invention significantly outperforms baseline methods in terms of rendering quality, pose accuracy, and physical plausibility.

This invention discloses a method and system for multi-view welding point cloud registration and stitching based on deep learning. The method includes: using a line-structured laser camera to scan the target weldment from multiple perspectives to obtain welding point clouds from different perspectives; obtaining the target point cloud corresponding to the target coordinate system; using a global registration network to obtain corresponding rigid transformation parameters based on the welding point clouds and target point clouds from different perspectives, and performing global registration on the corresponding welding point clouds based on the rigid transformation parameters to obtain the corresponding globally registered source point cloud; using a local registration network to obtain corresponding local transformation parameters based on the globally registered source point cloud and target point cloud from different perspectives, and performing local registration on the corresponding globally registered source point cloud based on the local transformation parameters to obtain the corresponding locally registered source point cloud; and performing downsampling processing on the point cloud of the overlapping region in the target coordinate system based on the locally registered source point cloud from different perspectives.

The application discloses a kind of based on initial calibration self-repairing DIC system field fast re-calibration method, device and system, and it is related to computer vision and optical precision measurement technical field.The application establishes unified space reference system by initial high-precision calibration, obtains camera internal parameter and external reference pose;When system occurs pose drift, field shooting calibration board image is adopted Global Constraint Optimization or single-point anchoring extremely simple strategy, and rigid transformation matrix is calculated to transform current coordinate system to initial reference;The transformation matrix is applied to all three-dimensional measurement data, and data unified correction is realized.The application does not need to reconstruct coordinate system, and it is simple and easy to operate, and correction is fast, can guarantee the spatiotemporal continuity and physical comparability of long-term monitoring data, and is suitable for large structure, precision equipment and other field long-term DIC measurement scene.Two kinds of strategies can be flexibly selected according to accuracy and efficiency demand.

The application discloses a kind of based on super deep well roadway three-dimensional measurement method of tank beam coordinate transmission, it is related to super deep well roadway measurement technical field, comprising: obtaining typical horizon tank beam endpoint and midpoint measured coordinates, and calculating unbiased baseline by linear interpolation;Offset peak horizon is identified based on the deviation of unbiased baseline;According to the height difference of target horizon and offset peak, peak weight is calculated, and three-point reference is obtained by correcting the amount of correction to modify unbiased baseline;Rigid transformation parameters are solved by using three-point reference to identify scanning target, and point cloud is converted to unified coordinate system.The application can improve the accuracy and reliability of super deep well roadway three-dimensional measurement.

The application discloses a local three-dimensional point cloud registration system and method, and the system comprises an encoder, a mapping module and SVD decomposition; the encoder module receives input point cloud data and image data, obtains a feature vector of the image data, performs feature extraction, and then transmits the features to the mapping module; the mapping module distributes the feature vector to a corresponding three-dimensional point cloud as an overlap score, then calculates a similarity matrix by using the calculated overlap score and the feature vector of the point, and then sends the calculation result to the SVD decomposition; the SVD decomposition receives the calculation result of the similarity matrix, and outputs a rigid transformation required for registration, and finally realizes local point cloud registration. According to the application, the key overlap feature points can be used to calculate a more accurate similarity matrix and rigid transformation, so that the local point cloud registration capacity is ensured to be accurate and fast in a low overlap scene and a complex multi-noise scene.

This invention relates to a point cloud registration method and system based on bounding box attention. The invention includes: acquiring point cloud data and generating source and target point clouds; performing feature extraction and multi-scale downsampling on the source and target point clouds to obtain superpoints and fine points, and establishing spatial associations between superpoints and fine points; inputting the superpoints into a GAT-based topological feature aggregation module to enhance their spatial topological features; introducing globally geometrically invariant features as cross-point cloud interaction priors through bounding box cross-attention, and performing spatial topological consistency verification of matching pairs using an AABB bounding box filtering mechanism to eliminate erroneous matches and obtain robust superpoint correspondences; propagating the superpoint correspondences to neighboring fine points to establish fine point correspondences; and finally estimating a rigid transformation matrix based on the fine point correspondences to complete point cloud registration. This invention effectively solves the technical problems of low accuracy, slow convergence speed, and poor robustness in low-overlap point cloud registration.

The application relates to an automatic general fracture simulation data generation method based on geometric parameter calculation, which comprises the following steps: obtaining a three-dimensional CT of a healthy target bone and a bone tissue binary label thereof, and constructing a feature distance graph representing bone structure and bone density; adopting a three-dimensional watershed algorithm to convert the feature distance graph into a plurality of independently connected bone fracture and fragmentation regions, wherein the three-dimensional watershed algorithm controls the change of the fracture degree by adjusting a water level super parameter; for each bone fracture and fragmentation region, a random rigid transformation is performed to generate a three-dimensional CT simulating a fracture. Compared with the prior art, the application has the advantages of parameterizable control and higher morphological fidelity of the generated simulation data.

The application provides a layered representation compression and progressive transmission method of Gaussian splat volume video. The layered representation compression method comprises the following steps: extracting scene surface geometry, constructing 4D Gaussian initial parameters as the basic representation of dynamic scene; calculating the importance measure of each Gaussian primitive, and dividing the Gaussian into L layers of progressive details; performing inter-frame modeling on the layered Gaussian through rigid transformation and residual compensation; calculating the average displacement of the Gaussian primitive, and dynamically adjusting the grouping length of the inter-frame modeling; injecting uniform noise simulation quantization error, and estimating the entropy of each attribute by using KDE; simultaneously constructing an end-to-end optimization weighted loss of distortion, code rate and time smoothness, and training the inter-frame modeling; and encoding the trained layered Gaussian parameters into 2D graphs and compressing by using H.264. The bit stream supporting multi-level quality selection is generated by the layered representation compression, the number of layers is selected for decoding according to the requirement, and progressive transmission and dynamic adaptation are realized.

The present application relates to the field of control, in particular to a kind of intelligent hoisting and sliding control method of pipe truss gas building modular unit.The content includes: the coordinates of sliding track support point and module bottom support point are measured, and rigid transformation of module coordinate system to construction coordinate system is solved, to obtain the target pose of module;The position of measurement target in construction coordinate system and module coordinate system is spatially registered, combined with the target pose of module, to obtain pose error;Based on the pose error, the target position of lifting point is calculated, and the correction displacement vector of lifting point in space is obtained, and then the length adjustment amount of sling is calculated, and the length of sling is adjusted;When module falls to the starting point of sliding track, introduce the track sliding distance control algorithm based on velocity integral constraint, control module sliding.The problems that the target pose determination of module at the starting point of sliding track is not accurate, the lifting point adjustment and sling length adjustment amount are difficult to accurately control in hoisting process, and the sliding distance control is not accurate are solved.

The application discloses a workpiece in-situ measurement method and device for grinding machine tool, and relates to the technical field of precision machining and measurement. The method comprises collecting an image sequence of a transverse section of a rotary workpiece and an image sequence of multiple stations along an axial direction; a one-dimensional gray profile is established along a normal direction for each pixel-level edge point in each frame of image; a parameterized fitting is performed on the error function model by introducing a linear background term, and a sub-pixel edge position is solved; a circle fitting is performed on a point set of the same height of the transverse section to reconstruct a transverse section line; a distance field is constructed for an axial profile point set of adjacent stations to establish a bidirectional Chamfer registration target function, and rigid transformation parameters are solved to realize accurate splicing of the axial profile; according to the reconstructed transverse section line and the axial profile line, a workpiece topography error is evaluated, and a detection result is output; the method realizes high-precision in-situ measurement of the transverse section and the axial profile of a large-size rotary part under a grinding environment.

The application provides a material handling robot automatic mapping and point calibration method. Applied to the technical field of material handling robots, the method comprises the following steps: acquiring scheduling system prior data, constructing a passable topological graph and a scanning constraint set, and generating an automatic scanning path; controlling the robot to collect multi-source sensing data along the path, fusing the mapping to obtain a target environment grid map; constructing a corresponding point set based on the scheduling coordinates and the map coordinates, calculating a rigid transformation matrix, and determining a global coordinate mapping reference after consistency verification; when the robot reaches the local exploration range of the target point, triggering local exploration to extract feature point clouds, obtaining pose deviation and exploration reliability; storing the deviation in the cache pool according to the reliability, calculating the point compensation value when the gradual stability condition is met, and completing the target point calibration. Thus, the accuracy of the material handling robot point calibration is improved.

The application discloses a preoperative CT and intraoperative X-ray film combined registration method and device, and the method comprises the following steps: obtaining preoperative CT data and intraoperative biplane X-ray film data of a target object; obtaining CT body data to be registered based on the preoperative CT data; obtaining intraoperative three-dimensional body data by pre-set three-dimensional reconstruction network reconstruction based on the intraoperative biplane X-ray film data; and registering the CT body data to be registered and the intraoperative three-dimensional body data by a pre-set three-dimensional registration network, and outputting six-degree-of-freedom rigid transformation parameters between the two to complete the registration. The application can realize high-precision and real-time combined registration of spine CT and X-ray films, output space registration parameters between images, and meet the precision and efficiency requirements in the image registration task.

The application discloses a kind of based on structure entropy sampling air-ground multi-source point cloud robust registration method, comprising: S1, respectively reading source point cloud and target point cloud, obtain the spatial coordinates, normal vector and curvature of source point cloud and target point cloud;S2, standardization pretreatment is carried out to source point cloud and target point cloud;S3, structure entropy sampling is carried out to source point cloud and target point cloud after standardization processing, and the sampling point set for registration is constructed;S4, the point-to-point registration method with robustness is used, the optimal rigid transformation matrix between source point cloud and target point cloud obtained by S3 structure entropy sampling is calculated, and point cloud registration is realized;S5, the registration result of S4 is restored in space, and the rigid transformation matrix under original coordinate is constructed;S6, output transformation matrix and registration point cloud after space restoration.The application is driven by structure entropy to sample and select key points, has the advantages that noise resistance ability is strong and registration precision is high, can satisfy the registration demand of small error air-ground fusion data.

The application discloses a large-span arch bridge digital pre-assembly manufacturing control method, relates to the technical field of bridge engineering digital manufacturing, and takes the semantic importance of components, the distribution density of point clouds and the design connection relationship as cores, constructs semantic weight, distribution weight and connection relationship weight, eliminates overall deviation through weighted centroid calculation and decentralization, fuses the weight to construct a covariance matrix and solves an optimal rigid transformation through SVD decomposition, realizes global optimization in combination with an improved graph cut algorithm, and finally outputs aligned point clouds containing manufacturing errors. The public segment is taken as a benchmark to calibrate a reference line shape, data support is provided for pre-assembly target line shape calculation of conventional segments, the alignment priority of the core area is effectively strengthened, sparse point cloud and local deviation interference is avoided, the overall splicing line shape consistency of arch ribs is ensured, and accurate technical basis is provided for segment splicing construction.

The application discloses a curved surface workpiece defect measurement method based on point cloud registration, comprising the following steps: obtaining a three-dimensional point cloud model and a theoretical point cloud model of a workpiece, and determining corresponding matching point pairs; based on the matching point pairs, a point cloud registration method based on graph optimization is used to rigidly register the two point cloud models, specifically comprising the following steps: based on the matching point pairs, a graph optimization model is constructed, the model regards each matching point pair as an edge in a graph structure, and defines a registration error as a multi-modal residual error including a geometric residual error, a normal vector residual error and a curvature residual error; an optimization objective function is constructed in combination with the multi-modal residual error and a robust kernel function, and a global optimal rigid transformation matrix is solved based on a graph optimization algorithm to realize point cloud registration; based on the point cloud registration result, a multi-modal residual error norm of each point is calculated to generate a residual heat map; and a defect area is identified through the residual heat map. The application realizes defect detection of a curved surface workpiece through multi-modal residual error registration of graph optimization.

This invention proposes a point cloud registration method and system for power battery reconstruction, relating to the field of 3D point cloud data processing technology. The method includes: establishing an initial point correspondence between a source point cloud and a target point cloud; using the PSO (Particle Swarm Optimization) algorithm to simulate and update the initial point correspondence, obtaining a coarsely registered first particle swarm; combining the ICP (Integrated Point Optimization) algorithm to minimize the Euclidean distance between corresponding points in the source and target point clouds, determining the optimal rigid transformation, and obtaining a finely registered second particle swarm; during each coarse and fine registration process, if the second fitness value corresponding to the second particle swarm is less than the first fitness value corresponding to the first particle swarm, the second pose of the second particle swarm is fused with the first pose of the first particle swarm to correct the first pose, and the target pose that minimizes the fitness function value is selected from the second pose to optimize the particle poses of the idle particle swarm of the first particle swarm, obtaining the target particle swarm.