22 results about "Vector angle" patented technology

The invention discloses a steel truss extraction method based on point cloud segmentation and feature analysis, and belongs to the technical field of three-dimensional point cloud processing. The method mainly comprises the following steps: firstly, carrying out multi-scale filtering denoising on an original point cloud and constructing an octree index; local geometric features such as normal vectors, curvatures and first-order plane fitting errors of all the points are calculated; based on the adaptively selected seed points, carrying out region growth by combining a normal vector included angle, a curvature difference and an adaptive edge criterion to realize preliminary segmentation; and finally, a coarse-to-fine multi-layer cascade classifier is adopted to screen the segmented areas, and the steel truss point cloud is accurately extracted through geometric dimension filtering, shape feature screening and multi-dimensional feature matching in sequence. The method does not need pre-training, has the advantages of strong anti-interference capability, high extraction precision and high calculation efficiency, and is suitable for engineering scenes of bridge detection, building surveying and mapping and the like.

The invention discloses a motion posture analysis method based on image recognition, particularly relates to the field of image recognition, and is used for solving the problem of force transmission posture recognition of rotation type throwing motion in the prior art. According to the method, an image sequence covering the whole throwing process is collected, a continuous displacement track from a pelvis to a wrist key point is tracked and extracted, a rotary force storage stage and a release output stage are divided based on key point dynamic behaviors, and an attitude conversion time boundary is constructed; extracting a rotation angle change sequence to generate an angular velocity curve, calculating a theoretical linear velocity based on a tail-section angular velocity and a connection vector length, establishing a velocity ratio by combining an actual throwing object initial velocity tracked in the image, and screening a sample with the highest energy conversion rate as an attitude structure reference; and finally, performing connection vector angle comparison with image sequences of other samples, and positioning a posture offset key point, thereby realizing identification and analysis of the throwing action power generation link deviation.

The present invention discloses an image matching method based on quadratic signature, including: obtaining a correspondence set; obtaining a first cost function based on the feature points and the neighborhood points of the feature points in the first relation set and the second feature point set; converting the first cost function into a simplified second cost function; obtaining a quadratic signature distance according to the weight vector between the feature points and the neighborhood points and the similarity matrix; obtaining a first quantization distance according to the consistency of the neighborhood topology of the vector angle and length, and obtaining a second quantization distance according to the consistency measure of the structure by the quadratic signature distance; converting the second cost function into a third cost function based on the first quantization distance and the second quantization distance; converting the third cost function into a fourth cost function, and minimizing the simplified fourth cost function to obtain an optimal correspondence set. The present invention can more accurately measure the similarity of local structures, improve the ability to remove incorrect correspondences, and retain more reliable correspondences.

The present application relates to a kind of alignment methods of the oblique angle of the oblique end face of pigtail, belong to optical fiber detection technical field, comprising the following steps: step 1) using ANSYS software adopts stereoscopic scanning method scanning and gathers the model data of pigtail and rectangular capillary;Step 2) the oblique angle that the end of pigtail needs to be ground is set in advance, the end of pigtail is simulated to be ground, and the deformation of pigtail grinding is analyzed;Step 3) using vector angle method analyzes the oblique angle of the end of pigtail after grinding, detects the oblique angle that the end of pigtail needs to be ground and the oblique angle of the end of pigtail after grinding;Step 4) the oblique angle of the end of pigtail after grinding is processed;The beneficial effects of the present application: by comparing the oblique angle that the end of pigtail needs to be ground and the oblique angle of the end of pigtail after grinding, the deviation between the oblique angle of the oblique end face of the pigtail after correction or detection grinding and the oblique angle of the oblique end face of the pigtail actually needed is reduced.

The invention discloses an analysis processing method based on epidermis texture graphic data, and relates to the technical field of epidermis texture data analysis, and the method comprises the steps: carrying out the three-dimensional scanning of a target epidermis to obtain an initial texture model, connecting three adjacent surface data points based on a Delaunay triangulation algorithm to form a triangulation analysis unit, and carrying out the triangulation analysis of the target epidermis. Calculating a normal vector included angle standard deviation of adjacent triangular patches in the triangular analysis unit where the texture reference point is located as a texture direction change rate, judging whether the texture direction change rate is greater than or equal to a preset threshold value, and performing skin curvature feature extraction on the current triangular analysis unit with first analysis intensity, according to the first analysis intensity or the second analysis intensity, epidermal frequency feature extraction is carried out on the current triangular analysis unit, so that different analysis intensities are intelligently switched, the key direction of feature extraction is flexibly adjusted, it is ensured that high-precision analysis is obtained in a key complex area, meanwhile, the overall calculation burden is effectively reduced, and the processing efficiency is improved.

This invention provides a training method and an evaluation method for a trajectory prediction model. The training method includes: obtaining the predicted trajectory of a target from the trajectory prediction model and dividing the predicted trajectory into multiple predicted sub-trajectory segments; obtaining the target's true trajectory and dividing the true trajectory into multiple true sub-trajectory segments, the number of true sub-trajectory segments corresponding to the number of predicted sub-trajectory segments; calculating the vector angles between the predicted and true sub-trajectory segments, including lateral and longitudinal vector angles; calculating the L2 distance loss between the predicted and true sub-trajectory segments; calculating the lateral and longitudinal distance losses based on the vector angles and L2 distance losses; and using the lateral and longitudinal distance losses as the loss function of the trajectory prediction model to optimize the parameters of the trajectory prediction model. This application performs a refined decomposition of the loss function, resulting in a more accurate prediction model after training.

The invention relates to a server and a method for processing a three-dimensional graphic file in a release format. The server is configured to analyze a vector included angle relationship between center point coordinates of all geometries corresponding to each three-dimensional model and center point coordinates of the corresponding three-dimensional model; comparing the vector included angle relations in the at least two three-dimensional models to preliminarily judge whether the spatial layouts of the at least two three-dimensional models are the same or not; under the condition that the layouts of the two three-dimensional models are the same, a plurality of geometries contained in the two three-dimensional models are subjected to one-to-one matching according to a preset matching rule, so that multiple pairs of matched geometries are obtained; and sequentially judging whether the to-be-compared geometries are consistent with one another according to the comparison sequence. According to the invention, the efficiency and accuracy of three-dimensional model matching search are greatly improved.

The application discloses a steel truss extraction method based on point cloud segmentation and feature analysis, and belongs to the technical field of three-dimensional point cloud processing. The method mainly comprises the following steps: firstly, performing multi-scale filtering denoising on original point cloud and constructing an octree index; then, calculating local geometric features such as normal vectors, curvatures and first-order plane fitting errors of each point; based on adaptively selected seed points, combining the normal vector angle, curvature difference and adaptive edge criterion to realize region growing and preliminary segmentation; finally, adopting a multi-level cascade classifier from coarse to fine to screen the segmented regions, and sequentially passing through geometric size filtering, shape feature screening and multi-dimensional feature matching to accurately extract the steel truss point cloud. The application has the advantages of strong anti-interference ability, high extraction precision and high calculation efficiency without pre-training, and is suitable for engineering scenes such as bridge detection and building surveying.

This invention discloses a method for measuring the vector angle of a thrust vector nozzle, comprising the following steps: S1, acquiring a high-speed schlieren image of the thrust vector nozzle wake to obtain an original grayscale image; S2, obtaining a preprocessed image and a set of coordinate points representing the main boundary of the wake; S3, obtaining the coordinate range of a first selected region and a second selected region; S4, filtering out a first set of edge points located within the coordinate range of the first selected region and a second set of edge points located within the coordinate range of the second selected region; S5, performing linear fitting on the first set of edge points to obtain a first fitted line, and performing linear fitting on the second set of edge points to obtain a second fitted line; S6, calculating the angle bisector of the two lines to obtain the equation of the angle bisector; S7, outputting the thrust vector angle. This invention enables the direct acquisition of the nozzle wake deflection angle based on a high-speed schlieren image in the measurement of the thrust vector nozzle wake deflection angle.

The application discloses a kind of multi-view point cloud matching feature histogram construction method based on structured light phase information, by introducing phase information as auxiliary feature, to enhance the feature distinguishing ability of fast point feature histogram (FPFH), first by structured light technology obtains phase information, combined with traditional fast point feature histogram algorithm, the extended fast point feature histogram feature descriptor including normal vector angle, Euclidean distance and phase value is constructed.Normal vector angle and distance are quantified to N interval respectively, and phase information is introduced as the second 2N+1 feature dimension, the distinguishing degree and robustness of feature description are significantly improved.In the feature matching process, pixel-level matching is carried out using polar correction and sum of absolute differences (SAD) algorithm, and sub-pixel optimization is carried out combined with the linear characteristics of phase information, and the matching accuracy is improved.The application is excellent in the point cloud registration of weak texture area and flat surface, effectively improves the quality and accuracy of three-dimensional reconstruction under complex scene.

The application relates to the technical field of deep learning, in particular to an image classification method, device and computer equipment for industrial detection, which can solve the problem of low classification detection accuracy when more fine sub-class division is performed on images due to small image difference. The image classification method for industrial detection comprises the following steps: obtaining a feature vector of a first training image, and inputting the feature vector into a weight matrix with multiple category vectors, wherein the category vectors are set according to image classification requirements; expanding each initial vector angle in an angle space to obtain a cosine value corresponding to a target vector angle, wherein the initial vector angle is an angle between the feature vector and each category vector; wherein the cosine value is used to determine a similarity, and the angle space is a classification space converted based on the weight matrix; and determining a first category corresponding to the first training image according to a corresponding threshold interval where the similarity is located.

The present invention relates to the technical field of pearl detection, and in particular to a pearl appearance quality detection method based on image processing. The method comprises the following steps: binarization processing; noise elimination processing; individual extraction processing; boundary extraction processing; boundary coordinate extraction processing; calculation of vector angle value mean; and classification and recognition. The method obtains image boundary coordinate value points, obtains vector angle values ​​composed of these points, calculates the mean value as a feature parameter, and inputs the mean value into a classifier to realize the recognition of the pearl appearance quality. Starting from the edge of the pearl, the feature parameter of the vector angle composed of the pearl edge points is found, thereby amplifying the edge information of the pearl and increasing the difference between pearls with qualified appearance quality and pearls with unqualified appearance quality. The method solves the problem that the concave or convex parts of pearls with unqualified appearance quality are relatively subtle and cannot be detected and recognized. The method not only improves the detection efficiency of the pearl appearance quality, but also reduces the false detection rate, thereby meeting actual detection needs.

The present invention discloses a method for correcting geometric attenuation factors based on optical microfacet theory. The method utilizes microfacet theory to assume that an object's surface is composed of multiple microfacets, each consisting of a first irradiation surface close to the incident light and a second irradiation surface farther from the incident light. Based on the geometric relationship between the first and second irradiation surfaces and the incident and reflected light, a normal vector angle probability model is employed to obtain a corrected geometric attenuation factor to determine the occlusion probability of the incident light. Substituting the corrected geometric attenuation factor into a bidirectional reflectance distribution function yields the radiant energy of the reflected light to detect the object's topography. This method can obtain an accurate and reasonable geometric attenuation factor, thereby enabling accurate analysis of the target's polarization characteristics and enhancing the effectiveness of polarization detection.

The present disclosure provides systems and methods for quantitative material characterization using multi-energy photon and neutron interactions are disclosed. Extending traditional dual-energy techniques, the disclosure utilizes multi-dimensional vector analysis from multiple energy bins to enhance material differentiation. The approach leverages distinct energy-dependent behaviors of photoelectric effect (PE), Compton scattering (CS), pair production (PP), Rayleigh scattering, and neutron interactions. By calculating differences, ratios, slopes, and vector angles and directions across energy channels, and constructing two-dimensional (2D) and three-dimensional (3D) vectors, unique material signatures are obtained. Vector angles and trajectories through quadrants correspond to materials like clock hands indicating time, demonstrating identification precision. The disclosure integrates neural networks trained on simulated data and incorporates real-time feedback loops correcting for dark current, noise, and detector drift before vector analysis. This eliminates Poisson and detector noise, ensuring vectors represent material signals matching simulation vectors.

The invention relates to the technical field of three-dimensional modeling, in particular to an automatic mapping method and system based on a three-dimensional model-engineering drawing, and the method comprises the following steps: sampling a three-dimensional model to obtain a vertex and a normal vector, calculating a curvature change rate through gradient operation, extracting features through principal component analysis to generate a reference set, and classifying detection value difference threshold values to form a region table; the method comprises the steps of generating a normal vector angle difference correlation area to generate a weight table, calculating a control point through vertex weighting and generating a coordinate set through projection, generating a result through Lagrange interpolation view alignment and layout adjustment, and generating a complete file through interpolation of a supplementary point set for checking coverage insufficiency. The interpolation function unifies multi-view-angle coordinates; the supplementary point set ensures continuous mapping; the output result is globally consistent and local integrity is considered; and gap offset is avoided to improve the mapping stability and integrity.

This invention relates to the field of 3D modeling technology, specifically to a bridge appearance defect detection system based on 3D point cloud reconstruction. The system includes: a point cloud spatial partitioning module, which constructs a 3D bounding box based on the bridge point cloud, calculates a local geometric degradation index, and compares the local geometric degradation index with a flatness threshold and a degradation threshold to obtain an initial 3D polygon mesh. In this invention, by filtering the normal vector angles of density feature patch regions and extracting polygon contours, local geometric abrupt change regions are separated from the overall structure. Furthermore, a dual filtering process is performed using the rate of curvature change and surface area values, ensuring that defect regions possess clear characteristic expressions in both spatial scale and morphological changes. This improves the accuracy of defect localization and ensures that bridge appearance defects remain identifiable even in complex environments.

The application provides a camera focal length and posture calibration method based on two straight lines, and belongs to the technical fields of computer vision, SLAM and photogrammetry. The application only needs to know two straight lines in space, measure the camera position and the two straight line positions, shoot straight line feature images, obtain the imaging relationship of the straight lines in the camera, utilize the important geometric phenomenon that the normal vector angles between two planes are equal, establish a method for quickly solving the camera focal length, and finally convert the corresponding relationship of the straight lines into the corresponding relationship of the space points between two coordinate systems with the same origin to solve the camera posture. Compared with the existing straight line-based calibration method, the application reduces the number of straight lines, shortens the calibration time, has a unique solution, and can simultaneously solve the focal length.

Systems and methods relate to the determination of accurate motion vectors, for rendering situations such as a noisy Monte Carlo integration where image object surfaces are at least partially translucent. To optimize the search for “real world” positions, this invention defines the background as first path vertices visible through multiple layers of refractive interfaces. To find matching world positions, the background is treated as a single layer morphing in a chaotic way, permitting the optimized algorithm to be executed only once. Further improving performance over the prior linear gradient descent, the present techniques can apply a cross function and numerical optimization, such as Newton's quadratic target or other convergence function, to locate pixels via a vector angle minimization. Determined motion vectors can then serve as input for services including image denoising.

The present invention discloses a double-rotation modulus calculation method with fast convergence, belonging to the technical field of wireless communication, and comprising the following steps: performing Fourier transform on the radar received signal to obtain a vector to be modulus-calculated; performing vector partitioning based on the accuracy requirement to obtain a plurality of vector regions; performing double rotation on the vector to be modulus-calculated, each vector region and the state machine to obtain the vector angle and the vector modulus value; in the present invention, by using the state machine to correspond to the vector region, after each double rotation of the vector, the absolute value of the ordinate of the rotated vector is re-compared to determine the current vector region of the vector, and the state machine is jumped to the corresponding state, thereby probabilistically skipping some states, reducing unnecessary rotations, and accelerating convergence. In this solution, the vector rotates two angles each time and is preferentially saved, which is equivalent to rotating at the optimal angle, can accelerate the convergence speed, and has higher accuracy when the input bit number is higher. The present invention solves the problem of excessive time consumption in the modulus calculation process of the LFMCW radar system.

The present invention discloses a multi-criteria extraction method for the surface trace of the jointed rock mass of a tunnel face based on a depth sensor, comprising the following steps: S1: obtaining point cloud data of the surface of the jointed rock mass of the tunnel face; S2: for any point in the point cloud, calculating the average value q of the sum of the normal vector angles between this point and all points in its neighborhood. i ; S3: Calculate the eigenvalues ​​of the covariance matrix of this point with respect to its neighborhood and calculate the flatness coefficient; S4: Based on the mean shift algorithm, preset bandwidth h i , perform iterative operations to obtain the convergence position, and calculate the distance D between this point and the convergence position i With bandwidth h i The ratio D i / h i S5: Calculate the characteristic value of this point, determine whether this point is a trace feature point, and filter out all trace feature points; S6: Use the trace feature points to extract the trace contour line. This solves the problem that traditional joint trace extraction methods are single, low in precision, and too costly.

A method for calculating an angle of the inductor flux vector, comprising a plurality of iterations, each iteration of the plurality of iterations except the first comprising: a / obtaining (S100) input data including armature voltage values, armature current values, armature resistance values, and armature inductance values; b / estimating (S101) the inductor flux vector of the iteration by adding to an inductor flux vector obtained in the previous iteration a function of said input data; c / a correction comprising i. determining (S102) the center of a circle passing through at least three points defined by the flux vector obtained by said estimation, and at least two flux vectors obtained in two preceding iterations; and ii. obtaining the flow vector of the current iteration in which we correct (S106, S107) the flow vector obtained by said estimation using said center vector.Figure for the abridged version: [Fig. 2].