578 results about "Euclidean distance" patented technology

The invention relates to the technical field of tunnel and underground space robot autonomous navigation, and provides a tunnel path planning method and system for clearance corridor and elastic geodesic refinement, and the method comprises the steps: calculating the Euclidean distance from a structural terrain to a nearest obstacle or wall, taking the Euclidean distance as clearance, generating a direction-independent cost, and constructing a dual metric; running Riemannian geometry fast travel and backtracking geodesic under the condition that starting and ending point pixel coordinates are given to obtain a coarse path, after rasterization, calculating the distance from pixels to the coarse path, obtaining a widening radius by using clearance along the coarse path, performing nearest neighbor propagation to a global domain to obtain a radius graph, defining a corridor, and performing corner propagation by the coarse path to obtain an approximate curvature field; calculating to obtain a score, partitioning through thresholding, performing piecewise linear penalty, updating the total cost related to the direction, re-marking according to partitions, taking the reciprocal as a velocity field, and obtaining a final path through a solver. The problems of safe obstacle avoidance, curvature control and efficient solution need to be met at the same time in engineering are solved.

The invention discloses an electromagnetic field simulation grid adaptive generation method based on a neural network. The method comprises the following steps: performing mesh generation on a semiconductor device simulation model to obtain original generation data, and constructing node features; an undirected graph is constructed, graph nodes of the undirected graph adopt node features of grid points, and connecting edges adopt Euclidean distances between the grid points and adjacent grid points; inputting the undirected graph into a double-branch neural network, and outputting a predicted node position; and correcting each node of the original subdivision data by using the predicted node position obtained by the double-branch neural network to form new subdivision data for electromagnetic field simulation. According to the method, physical field information can be fused to efficiently adjust the node density, and the geometric boundary and topological structure characteristics of the device can be strictly kept.

The invention provides a low-airspace public route construction method, device and equipment, and relates to the technical field of aircraft airspace route planning. The method comprises the following steps: determining a plurality of target candidate nodes corresponding to a city based on ground traffic topological characteristics of the city corresponding to a low airspace and spatial distribution characteristics of low-risk ground features; performing triangulation on the plurality of target candidate nodes to obtain an initial edge set; wherein the initial edge set comprises a plurality of initial edges; performing sparse processing on the plurality of initial edges based on a greedy tensor graph algorithm, and determining a plurality of target edges from the plurality of initial edges; wherein the length of the shortest path between the two nodes of the target edge is greater than the greedy tensor coefficient times of the Euclidean distance between the two nodes; and constructing a low-airspace public route based on the plurality of target edges. By adopting the technical scheme provided by the invention, the utilization efficiency and safety of the low airspace can be effectively improved by constructing the unified low airspace public route.

The invention discloses a mobile robot path planning method based on an improved A star algorithm, which comprises the following steps of: firstly, acquiring environment information of a current map through a sensor, establishing a grid map, and simultaneously determining a starting point coordinate, an end point coordinate and an environment obstacle coordinate of a robot; an OPENLIST list and a CLOSEDLIST list are created, and a starting point is put into the OPENLIST list; selecting a new distance calculation function, and setting the Euclidean distance as a median value of the Euclidean distance and the Manhattan distance; improving a heuristic function, and introducing an adaptive estimation function; the total distance cost of the nodes is calculated from the starting point according to the node expansion mode, the optimal path nodes are placed in the CLOSEDLIST till the end point is traversed, and the optimal path is generated; and deleting redundant nodes in the path based on a collinear principle to obtain a final optimized path. The search time of the algorithm is shortened, the algorithm can better adapt to different environments and scenes, the applicability of the algorithm is enhanced, and the planning efficiency of the algorithm is effectively improved.

The invention relates to the technical field of image recognition, in particular to a plastic particle black spot defect online detection method based on deep learning. Performing non-linear space conversion on the original image, and separating a background field component and a significance component; constructing a reverse compensation item by using the background field component, carrying out dynamic gain correction on an original brightness channel, and carrying out multi-channel weighted fusion on the original brightness channel and the saliency component to generate a saliency feature map; inputting the saliency feature map into a multi-scale topology enhancement network, extracting edge distribution features, and performing closed contour extraction and Euclidean distance transformation to generate a topology thickness energy map; the high-frequency gradient magnitude of the original image is calculated, mask smoothing processing is carried out on the gradient of the black spot region by using the saliency feature map, and a joint dissipation field is constructed; and taking a local maximum value point in the topological thickness energy diagram as a morphology seed point, executing controlled watershed evolution under the topological constraint of the joint dissipation field, realizing boundary stripping of an adhesive particle region, and generating a particle morphology diagram.

The invention discloses an electric power equipment safety long text knowledge retrieval method based on sparse and dense joint enhancement, which comprises the following steps of: encoding an electric power equipment safety long text and retrieval query by utilizing a pre-training language model for performing supervision and fine tuning on an electric power equipment safety field label data set, and generating a semantic word vector; a weighted sparse vector fusing semantic information and a dense semantic vector enhanced through nonlinear transformation are generated through fine-grained semantic enhancement branches; respectively calculating sparse similarity based on a common item weight product and dense similarity adopting a double-distance weighted fusion strategy combining a cosine distance and an Euclidean distance; and carrying out weighted summation on the two similarity scores through the learnable weight to obtain a final similarity, and realizing accurate sorting retrieval of the long text knowledge. The method gets rid of dependence on global features of sentence vectors, can accurately capture local key information in a long text, and has high retrieval precision and semantic robustness.

The invention provides a structured light projector calibration method and system and a medium, and the method comprises the steps: projecting a multi-frequency stripe pattern to a target region based on an MEMS micromirror array module, obtaining a distorted image sequence, and carrying out the phase decoding, and obtaining a phase diagram corresponding to each frequency; resolving the phase diagram based on a phase shift method and an optical path triangulation principle to obtain a preliminary three-dimensional point coordinate; constructing a neural network model and a calibration sample set, obtaining a real three-dimensional coordinate by using a standard calibration plate or a known curved surface, and calculating the deviation between the initial three-dimensional point coordinate and the real three-dimensional coordinate to obtain a training sample set; and training the neural network model according to a training sample set based on an Euclidean distance loss function, outputting a three-dimensional offset compensation amount according to the neural network model, correcting the initial three-dimensional point coordinates, and realizing an end-to-end error self-compensation mechanism through the neural network model, so that deviation compensation is effectively performed on the three-dimensional point coordinates, and the accuracy of the three-dimensional point coordinates is improved. And the three-dimensional reconstruction precision is improved.

The invention relates to an airspace intelligent generation planning method, system and device based on live-action three-dimensional and semantic rules and a medium. The method comprises the steps of performing Euclidean distance field calculation on a target area live-action three-dimensional triangular mesh model, generating a three-dimensional voxelization distance field, extracting a continuous curved surface formed by points with distance values equal to a safety radius in combination with the safety radius of an aircraft, and generating a flight channel; performing airspace constraint rule mapping processing in the flight channel to generate a flight space with a cost attribute; path searching is carried out in the flying space, an initial broken line path is obtained, spline curve fitting processing is carried out on the path, and a continuous flying track is obtained. According to the method, modeling is carried out on geometric information of a live-action three-dimensional model and multi-dimensional airspace constraints, and path searching and trajectory fitting are combined, so that the environment authenticity, path safety and adaptability to aircraft dynamic characteristics of airspace planning are improved, and the responsiveness to dynamic airspace changes is enhanced.

The invention belongs to the technical field of mechanical fault diagnosis, and discloses a rolling bearing small sample fault diagnosis method and system based on an online soft label Gaussian prototype network. The method comprises the following steps: collecting bearing vibration signals under different working conditions to construct a data set, dividing the data set into a meta-training set and a meta-test set, and splitting the data set into a support set and a query set; constructing a Gaussian prototype network model containing an embedding module, a prototype calculation module and a classification module; in the meta-training stage, soft labels are dynamically generated by adopting an online soft label strategy, and multi-task training is carried out in combination with hard label loss to obtain optimal parameters; in the meta-test stage, a Gaussian prototype is constructed based on a support set, classification is achieved by calculating the Euclidean distance between a sample and the prototype, and prototype parameters are finely adjusted to adapt to cross-working-condition diagnosis when feature distribution drifts. According to the method, label noise interference is effectively relieved, the diagnosis precision can still be ensured under a small number of labeled samples, and the model generalization ability and the diagnosis stability are remarkably improved.

The invention relates to a basketball player motion capture analysis method and system based on an industrial camera. The method comprises the following steps: based on a preset three-dimensional human body model, obtaining a three-dimensional human body surface grid sequence and internal skeleton joint kinematics data by combining a pixel-level mask of each multi-view image frame set; detecting a body contact event and a continuous time period based on the Euclidean distance of the three-dimensional human body surface grid sequence; based on the internal skeleton joint kinematics data, calculating a multi-dimensional kinematics feature time sequence; constructing a dynamic causal hypothesis based on the body contact event, checking the dynamic causal hypothesis based on a preset differentiable forward dynamic multi-rigid-body human body model and a multi-dimensional kinematics characteristic time sequence to obtain a loss function value, and judging the hypothesis with the smaller loss function value as a real dynamic causal relationship. And the identity identification information is combined to obtain the identity of the active force generation party and the identity of the passive force bearing party. By adopting the method, a real force exerting main body and a real force bearing main body in confrontation can be accurately identified.

The invention provides an industrial anomaly classification method and system based on a multi-modal large language model, and belongs to the technical field of visual detection. The visual pre-screening module adopts a BYOL unsupervised feature extractor and a local block matching technology to calculate the minimum Euclidean distance between an image to be detected and a normal sample so as to generate an abnormal thermodynamic diagram, and efficient preliminary screening is realized; the coordinate set is converted into a binary mask through an abnormal contour marking module, and an accurate single-pixel contour is generated through morphological gradient and SIFT displacement correction; fusing a CLIP visual encoder and a GPT text encoder in a multi-modal decision engine, realizing visual-semantic alignment through cross-modal fusion and a multi-head attention mechanism, and supporting alpha hybrid dynamic rule update; and a dynamic classification executor calculates an abnormal region risk index based on the risk grading model, and generates a grading response action and a structured report. According to the invention, the classification precision is improved, so that the method can adapt to complex and diversified industrial abnormal scenes.

The invention relates to the field of computer data processing and artificial intelligence, and discloses a mobile phone APP intelligent code scanning and AI automatic identification system which comprises a mobile terminal, an internet network and a local server. The mobile terminal is provided with an acquisition detection module, and a mobile phone end code scanning module automatically controls light supplement when the image brightness is lower than a threshold value. A server-side AI intelligent recognition module executes Gaussian filtering, binarization and geometric correction preprocessing on a code scanning image, and a text is extracted by using a deep convolutional neural network and a bidirectional long-short-term memory network. And the system further executes semantic analysis by traversing the business template library, judges the attribution of key fields by using a spatial Euclidean distance, and generates structured JSON data. The statistical module calculates a real-time production efficiency index, and the remote viewing module associates and displays a process optimization suggestion when the data is abnormal. According to the invention, high-precision identification, business semantic understanding and intelligent decision support of code scanning data are realized.

The invention discloses a daily load curve clustering method based on a mixed distance improved FCM algorithm, and belongs to the technical field of data processing, and the method comprises the steps: collecting daily load data of a user at a sampling frequency of 5 minutes, carrying out the corresponding data preprocessing, analyzing the fluctuation characteristics of a daily load curve, and carrying out the clustering of the daily load curve; euclidean distance representing load curve power amplitude information and cosine distance representing load curve power fluctuation direction information are constructed respectively, and then an entropy weight method is adopted to construct a hybrid similarity distance (HSD) fusing the Euclidean distance and the cosine distance as a similarity criterion between daily power load curves; replacing the Euclidean distance in the FCM algorithm with the constructed HSD, then calculating a corresponding membership degree and a clustering center, iteratively updating the membership degree and the clustering center by using a target function based on an intra-class error weighted quadratic sum, and completing daily load curve clustering according to a preset maximum iteration number; and finally, determining the optimal clustering number and the corresponding optimal daily power load curve clustering result by using the sum of intra-class errors (SAE), thereby effectively solving the prominent problems of poor effect, low efficiency and the like of the traditional power load curve clustering method.

The invention relates to a clustering method for recognizing a semantic and structural relationship of text data based on a granular ball model, and aims to solve the problem that semantic association, spatial distribution and boundary sample label distribution deviation are difficult to consider by single measurement in high-dimensional text clustering. The method comprises the following steps: constructing a text data set into initial pellets and adding the initial pellets into a set; bisecting the split pellets based on linear perception measurement, and screening high-quality pellets through a weighted DML value; secondary splitting is carried out according to the Euclidean distance, and particle ball distribution is optimized according to a weighted DME value; performing supplementary splitting on the overlarge pellets to obtain a final pellet set; using a K-means algorithm to cluster the particle ball center points to generate a center set; and redistributing all samples to the nearest clustering center, and outputting a result. According to the method, two measurement modes are fused, text linear semantics and spatial structure characteristics are accurately captured, boundary sample label distribution is optimized, meanwhile, the calculation scale and parameter tuning complexity are reduced, semantic information loss is reduced, the precision, stability and efficiency of high-dimensional text clustering are remarkably improved, and the method is suitable for being popularized and applied. The method is suitable for multiple scenes such as information retrieval and theme mining.

The invention relates to the technical field of color sorters, in particular to a color sorter material recognition system with a self-learning function, which comprises a scene adaptation module, a hierarchical self-learning module, an interpretable decision module, a man-machine interaction module and a data storage module, and the scene adaptation module is electrically connected with the scene feature library, is matched with a preset scene feature library and outputs model initial adaptation parameters, and the hierarchical self-learning module is electrically connected with the scene adaptation module and receives the model initial adaptation parameters. Through multi-sensor environment acquisition and scene feature library matching of the scene adaptation module, rapid adaptive adjustment of initial parameters of the model is realized, and adaptive parameters are generated through weighted Euclidean distance algorithm matching or interpolation, so that the initial recognition accuracy of the model in a new scene is improved, the problem that the cross-scene performance of the model is suddenly reduced in the prior art is solved, and the recognition accuracy of the model is improved. And the flexible production requirements of multiple producing areas and multiple factories are met.

The invention relates to the technical field of oil and gas reservoir development and geological modeling, and discloses a multi-scale factor constraint-based tectonic network fracture modeling method, which comprises the following steps of: synthesizing a Riemannian metric tensor field on the basis of earthquake, logging and geomechanics data, and defining non-Euclidean distance cost of a fracture expanded in an anisotropic medium; initial seed points are screened according to the elastic strain energy density, and initial growth potential energy in a limited range is distributed; solving the eikonal equation by using an anisotropic fast marching algorithm to carry out wavefront competitive growth, and dividing a grid region into generalized Voronoi units; identifying a wavefront contact interface, and extracting gradient features to judge a fusion or truncation type so as to establish fracture topological connection; and finally, tracking a geodesic line path along an anti-gradient direction to generate a three-dimensional discrete fracture network. According to the method, macro and micro constraints are unified through Riemannian geometry, clear physical significance is given to the fracture by utilizing an energy mechanism, and automatic and accurate construction of the complex fracture network topology structure is realized.

The invention discloses a multi-task fixation point estimation method, which relates to the technical field of computer vision and deep learning, and comprises the following steps of: firstly, constructing a data set for providing associated data from tasks to labels and supporting multi-task cooperative training, and then constructing an MTHGaze model, through a semi-structured framework, deep learning, a geometric constraint fitting model and key tasks of cooperative processing are fused, a loss function is designed and trained, defects of each task are solved through targeted loss function design, and a scientific training process enables the model to be iteratively optimized. Finally, the targets of improving the estimation precision of the fixation point, ensuring the physical rationality and enhancing the environmental robustness are achieved; and finally, performing fixation point mapping. According to the estimation method, through data set construction, data consistency can be guaranteed, training noise can be reduced, a real reference can be provided for loss function calculation, error back propagation can be realized, and Euclidean distance errors can be minimized through L2 norms of real fixation point coordinates and prediction coordinates.

The invention discloses a method for measuring the thickness based on laser profile scanning of a 3D wheel of a conveying belt, particularly relates to the field of laser scanning thickness measurement, and is used for solving the problems that shutdown is needed and the measurement range is limited in existing conveying belt thickness detection. According to the method, three-dimensional point cloud data of the upper surface and three-dimensional point cloud data of the lower surface are collected at the same time, initial segmentation of the upper surface and the lower surface is completed through double criteria based on curvature and normal, a dynamic stability threshold value is constructed in combination with the running speed and transverse swing of the conveying belt, and a stable surface point set is formed through screening; then, a thin-plate spline algorithm of dynamic tensor weight is adopted to respectively fit continuous upper surface and lower surface reference curved surface models; and finally, the sampling points on the upper surface are projected to the lower surface along the normal direction and the Euclidean distance is calculated, so that the real thickness distribution of the conveying belt in the full width and full length range is obtained. Non-stop, non-contact and high-coverage online thickness measurement is realized, and the operation safety and maintenance management efficiency of the conveyor belt are improved.

The invention discloses a calculation method for integrating task induction and intrinsic spontaneous brain function activity. The calculation method comprises the following steps: calculating a brain activation mode when an individual executes a corresponding cognitive task based on task state functional magnetic resonance imaging data and a general linear model; identifying individual large-scale nerve avalanche with spatial continuity based on resting state functional magnetic resonance imaging data; the method comprises the following steps: performing principal component analysis on resting state functional magnetic resonance data of an individual to construct a low-dimensional state space; a task-induced brain activation mode and intrinsic spontaneous nerve avalanche are projected to an individual low-dimensional state space; calculating the Euclidean distance between the task-induced brain activity and the intrinsic spontaneous nerve avalanche in the low-dimensional state space; and detecting the prediction effect of the geometric distance on the performance of the tested task through the regression model. The method is verified on a real data set, and experimental results show that the method not only can integrate two basic brain function activities, but also can significantly predict individual cognitive performance differences.

The invention relates to an IEC61499 intelligent model segmentation training method based on DCU state enhancement, and belongs to the technical field of computer distribution. The method comprises the steps of constructing an intelligent model state enhancement module, transplanting an intelligent model to a DCU accelerator card, adjusting the number of attention heads through state vector complexity, carrying out parallel calculation, fusing and optimizing a multi-head attention matrix, sampling an optimal historical state based on Euclidean distance, and obtaining an enhanced state sequence in combination with a current state. Providing optimization state representation for segmentation learning; taking the enhanced state sequence as initial input, constructing and training a reinforcement learning agent, and enabling the reinforcement learning agent to learn and recognize an optimal segmentation strategy of the intelligent model; and deploying the trained intelligent agent, and performing intelligent splitting on the intelligent model according to the calculation load and the communication cost between the acceleration cards to realize distributed parallel training. The invention aims to solve the technical problems of unbalanced resource allocation, low model segmentation efficiency and redundant fixed attention mechanism calculation in a dynamic environment in the prior art.

A flood similarity intelligent analysis method based on multi-mode Transform and comparative learning comprises the steps that firstly, static features are processed through a grouping full-connection network, and the characterization capacity is enhanced through feature specific transformation; a CNN-Transform-Attention hybrid network is constructed to extract dynamic process line features, a convolutional layer captures local morphological features, a Transform encoder captures a global time sequence dependency relationship through a self-attention mechanism, key hydrological stages are adaptively focused through an attention weight, multi-modal features are adaptively integrated by adopting a gating fusion mechanism to generate unified embedded representation, and the dynamic process line features are extracted by adopting a convolutional neural network (CNN)-Transform-Attention hybrid network. A comparison learning and difficult sample mining strategy is introduced, discriminative characterization is learned in a low-dimensional embedding space through a comparison loss function training model based on the Euclidean distance, and the similarity is mapped into an interpretable probability of a [0, 1] interval; the technical problem that multi-source heterogeneous features are insufficient in utilization is effectively solved, accurate quantification and interpretable evaluation of flood similarity are achieved, and technical support is provided for flood forecasting, historical flood matching and flood control dispatching decision making.

The invention relates to the field of data analysis, in particular to a farmland soil compaction diagnosis method based on multi-source sensor data, and the method comprises the steps: carrying out the gradient energy feature construction and phase alignment cross-correlation evaluation of soil resistance depth curve data, and obtaining a gradient energy phase locking factor; carrying out rheological consistency coupling evaluation on the morphological characteristics of the soil resistance depth curve and the soil volume moisture content to obtain a rheological morphological coupling factor; performing joint correction on the basic Euclidean distance to obtain an adaptive similarity distance and executing clustering iteration; spatial modeling and prescription mapping are carried out on a clustering result and sampling point position information, so that a soil compaction accurate diagnostic graph is obtained, and a differentiated subsoiling operation instruction is generated; the method solves the problem that the existing Euclidean distance easily causes the characteristic dislocation of the compaction layer and the deformation of the rheological strength under the condition of farmland micro-topography fluctuation and water content difference to cause the distortion of the clustering partition.

The invention relates to the technical field of video content analysis, in particular to a mine surveillance video key frame extraction method, which comprises the following steps: extracting multi-dimensional features of each frame of a candidate frame set, obtaining a fusion value, and obtaining a candidate key frame set; based on an Euclidean distance method, calculating an Euclidean distance between continuous frames of the candidate key frame set, and determining a total clustering number according to a relationship between the Euclidean distance and a preset clustering threshold value; and clustering the frames in the candidate key frame set by using a preset mixed GWO-FCM clustering algorithm and the total clustering number to obtain a key frame set. According to the method, a hybrid clustering algorithm GWO-FCM combining grey wolf optimization and fuzzy C-means is adopted, global search and local optimization capabilities are considered, and the accuracy and representativeness of key frame clustering are ensured. The high-quality extraction of the key frames is realized, the number of redundant frames is obviously reduced, and the compactness and readability of the video abstract are improved.

The invention discloses a deep learning point cloud registration method for low-overlapping-rate point clouds. The method comprises the following steps: (1) carrying out multi-scale down-sampling on an input point cloud, calculating a weighted geometric centroid and a normal vector based on an adaptive neighborhood, and constructing a stable local reference point; (2) proposing a centroid geometric structure code, and embedding an attention mechanism by taking an inter-centroid Euclidean distance, a normal included angle and a neighborhood radius ratio as geometric priori to realize feature matching of structure constraint; (3) in a training stage, optimizing overlapping region positioning and point pair matching consistency by adopting a joint loss function formed by circular ring feature loss, overlapping perception loss and matchability loss; and (4) rigid body transformation is estimated by combining a coarse-to-fine matching strategy with a random sampling consistency algorithm, and global high-precision alignment is realized. According to the method, the registration precision and robustness under the conditions of low overlapping rate, noise and shielding are effectively improved, and the method is superior to an existing method in the aspects of a standard data set and real measurement data.

The invention discloses an overhead line fault diagnosis method, system and device based on space vector conversion and a storage medium, and relates to the technical field of intelligent power distribution, and the method comprises the steps: collecting the normal operation three-phase current data of an overhead power distribution line, and constructing a steady-state feature model; three-phase current data is collected in real time and converted into a space vector sequence, the Euclidean distance between a vector end point and a model center is calculated, and instantaneous offset is extracted. Integrating the offset in a sliding window, calculating a phase rotation rate, generating a multi-dimensional deviation tensor, inputting the multi-dimensional deviation tensor into a fusion function to obtain a misconstruction score, and normalizing the misconstruction score to obtain a misconstruction index; comparing the index with a threshold value, if the index exceeds the threshold value, determining that the disturbance is external transient disturbance, and adjusting the sampling frequency of the current sampling device; according to the method, interference can be effectively identified in a special electromagnetic environment, misjudgment of diagnosis is avoided, maloperation of a protection device is reduced, reclosing operation is guaranteed, and the fault diagnosis accuracy and robustness of a power distribution system are improved.

The invention discloses an FPGA point cloud registration acceleration method and system based on a scanning line auxiliary distance projection structure, and the method comprises the steps: carrying out the projection of a source LiDAR point cloud according to a distance and an azimuth angle, taking the Euclidean distance from a point to a sensor in an azimuth angle partition as a distance domain, dividing the distance domain into distance scale segmentation domains, and distributing distance scales, counting points in a distance scale segmentation domain, accumulating according to the distance scale to obtain an initial position, forming an SA-RPS-Index initial index array, executing counting sorting on source LiDAR point clouds, rearranging points which belong to the same distance scale segmentation domain and are positioned in the same distance scale into a continuous memory, generating an SA-RPS-Point point array, and storing the SA-RPS-Point point array in the SA-RPS-Index initial index array in the SA-RPS-Index initial index array in the SA-RPS-Index initial index array in the SA-RPS-Index initial index array. A scanning line auxiliary distance projection structure is obtained in combination with an SA-RPS-Index initial index array, each query point in a target LiDAR point cloud is marked to extract a candidate point, the distance between the candidate point and the query point is calculated, a PNN-C triple and an ENN-C edge corresponding point set are obtained in combination with K neighborhood and scanning line attributes, rigid body transformation between a source point cloud and a target point cloud is solved, and the target point cloud is obtained. And obtaining a motion estimation result.

The invention discloses an intelligent driving recognition system and method based on deep learning, and relates to the technical field of automatic driving and intelligent cockpit display, the system comprises a data acquisition and providing module, a neural network model module, a data post-processing and fusion module, a graphic rendering and control module and a display module, and RGB images and three-dimensional point clouds are acquired, so that the three-dimensional point clouds can be obtained; the method comprises the following steps: executing pixel-level semantic segmentation by using a hybrid neural network combined with a residual block, generating a drivable area and an obstacle mask, performing coordinate registration on point cloud by a system, screening and aggregating obstacle point cloud clusters by using a segmentation result, executing time sequence filtering processing on a road boundary based on an Euclidean distance, and generating stable boundary information. A graphics rendering module generates a geometric model accordingly. According to the invention, the problems of boundary jump identification and non-standard obstacle rendering distortion are solved, the display of driving, parking and navigation functions is realized, and the precision and interactive experience of intelligent driving perception are improved.

The invention provides an image restoration method and system based on reverse cascade energy structure co-evolution, a medium and equipment, and belongs to the technical field of image processing and image restoration. The system comprises: a front-end CNN configured to perform visual feature extraction to obtain an initial feature map; the distance map generator is used for calculating the shortest Euclidean distance from each pixel in the damaged area to the known area to obtain a distance map; the damaged area is divided into equidistant concentric annular belts; the boundary statistical encoder is used for calculating a statistical moment of each annular belt and a joint feature of each pixel in each annular belt; the reverse band cascade HDNN is used for calculating the Hamiltonian amount of each annular band by using a Hamiltonian function and executing half-step symplectic evolution according to a sequence from an outer band to an inner band, and the Transform repair network is used for repairing a damaged area by using the Transform repair network by taking an evolution result as input and outputting a repaired image. According to the scheme, the structural consistency of damaged image restoration can be ensured, distortion is avoided, and the restoration effect is improved.

The application belongs to the technical field of electricity, and particularly relates to a large language model driven conditional diffusion new energy output scene generation method. First, a conditional diffusion model suitable for new energy output scene generation of the power grid side is constructed, the conditional probability distribution of the actual output of new energy is learned implicitly by embedding the conditional information, and a scene set is generated based on a Markov chain. Second, a thinking chain optimization framework driven by a large language model is proposed, the semantic reasoning capability of the large language model is used to analyze the training state, and the optimal parameter interval is quickly locked under extremely low computing budget. The example verification result shows that the average Euclidean distance of the generated scene set is improved by more than 30.7% compared with the traditional method; the optimization efficiency and timeliness are greatly improved, under the extremely low computing budget of only allowing 10 iterations, the scene generation accuracy is further improved by about 1% compared with the Bayesian optimization method, and the scene set generation time is reduced by 87% compared with the Copula model; the reliability and timeliness of the dispatching decision are significantly improved.

The application relates to the technical field of machine vision and image processing, and discloses a plastic valve visual defect real-time detection method and system based on deep learning, which comprises the following steps: acquiring a gray image containing a plastic valve, positioning the centroid of a plastic valve main body area by using geometric moments; converting the gray image to a polar coordinate based on the centroid, applying one-dimensional high-pass filtering, and determining a dynamic segmentation threshold to extract abnormal gray pixel points; performing spatial density clustering after reverse mapping to determine a suspected defect area; calculating the Euclidean distance between the centroid of the suspected defect area and the centroid of the plastic valve main body area, synchronously inputting the distance and image features into a double-input convolutional neural network to output a plastic valve defect detection result. The application filters out cloud and fog-shaped diffuse reflection interference, and improves the accuracy of defect detection.