30 results about "Hilbert curve" patented technology

The invention discloses a serialization-based point cloud over-segmentation method, device and equipment and a medium, and relates to the technical field of point cloud over-segmentation. The point cloud over-segmentation method comprises the following steps: obtaining point cloud data set preprocessing; serializing the preprocessed point cloud data into a Hilbert curve; dividing the serialized point cloud data into a plurality of initial segments based on spatial continuity, and obtaining a multi-scale segment set; the feature similarity between adjacent segments is calculated, and a similarity matrix is obtained and established to provide a basis for subsequent super-point clustering; carrying out super-point clustering on the initial segments based on a self-adaptive updating algorithm; updating super-point features through a cross attention mechanism, establishing dynamic association between points and super-points, and obtaining a new super-point structure; and according to the updated super-point features, constructing a super-point graph, performing enhancement processing through a graph convolutional network, fusing features from the backbone point cloud network and corresponding multi-level super-point features, then transmitting the fused features to a segmentation head for segmentation, and obtaining semantic segmentation output.

The invention discloses a point cloud registration method based on an attention mechanism and a state space model. The method comprises the following steps: obtaining a standard surface shape precision point cloud and a deformed surface shape precision point cloud of an astronomical instrument; carrying out different levels of downsampling processing on the point cloud; sequentially carrying out serialization of a Hilbert curve, a transposed Hilbert curve and a zero-order curve on the processed point clouds of different levels; inputting the serialized point cloud into a coarse registration network fusing an attention mechanism and a state space model to extract features, and obtaining a corresponding relation of point cloud features; and estimating initial point cloud rotation and translation transformation through a dense-sparse fine registration network based on the corresponding relationship, and finally optimizing initial transformation through a weighted singular value decomposition algorithm to obtain a final registration result. According to the point cloud registration method, the registration effect of the point cloud registration method in a large-scale precise point cloud scene is improved, good registration performance is also ensured, and the effect of point cloud registration in surface type detection of high-precision equipment such as astronomical instruments is greatly improved.

This invention provides a method, apparatus, device, and medium for UAV route planning based on Hilbert curves. The method includes: collecting real-time position, attitude, and environmental information of the UAV; constructing an initial UAV route using a Hilbert curve based on the UAV's mission requirements and the collected information; smoothing the initial UAV route and removing redundant paths using a Bézier curve to obtain an optimized UAV route; correcting the optimized UAV route based on real-time environmental changes to generate a safe route; and converting the safe route into executable control commands for the UAV to drive its operation, thereby solving the problems of insufficient space filling efficiency, poor adaptability to dynamic environments, and low path smoothness in dynamic path planning.

The application discloses a non-full-field data synchronization method based on a Hilbert curve and related equipment thereof, and belongs to the technical field of big data. The application obtains the primary key information of non-full-field data, performs a hash operation on the primary key information to obtain a hash value corresponding to the primary key information, maps the hash value to a two-dimensional space where a Hilbert curve of full-amount data is located, obtains coordinate data of the hash value on the Hilbert curve, inputs the coordinate data into a preset Hilbert space curve algorithm model to obtain position information of the non-full-field data, obtains a synchronization mode corresponding to the non-full-field data, and synchronizes the non-full-field data based on the synchronization mode and the position information. In addition, the application also relates to the technical field of blockchains, and the non-full-field data can be stored on nodes of a blockchain network. The application can realize fast query and positioning of non-full-field data, quickly complete synchronization of non-full-field data, and improve the efficiency of data synchronization.

The application belongs to the field of computer application, and discloses a directed road network error limited Hilbert tree learning index construction method, which projects a space-time sample into a position-time grid, sorts cells of the grid based on a Hilbert curve, then divides the sorted sample into logical blocks and stores the logical blocks into a disk, and simultaneously establishes a binary tree for dividing nodes according to Hilbert curve serial numbers, representing a mapping relationship between the Hilbert serial numbers and the logical blocks, and ensuring that the mapping error of leaf nodes is less than a specified error. The application divides a two-dimensional space according to the Hilbert curve, ensures that trajectory data in the position-time grid is uniformly distributed, ensures full recall of query data through bounded error, and thus improves the query performance on inclined data.

This invention discloses a liquid-cooled plate flow channel structure based on Hilbert curves, belonging to the field of power battery thermal management technology. It mainly involves a coolant flow channel network composed of multiple flow channel units. The coolant flow channel network includes multiple integrated main channels, each consisting of multiple flow channel units connected in series. Each flow channel unit includes a unit inlet end, a Hilbert curve flow channel, and a unit outlet end. The liquid-cooled plate has inlets on both sides of its long side and outlets on both sides of its short side, forming intersecting transverse flow channels. This invention utilizes the self-similarity and high spatial coverage of the Hilbert curve to achieve near-full coverage flow of the coolant, expanding the heat dissipation area. Furthermore, it enhances heat transfer through continuous turns and flow channel contraction-expansion, significantly improving temperature uniformity and heat dissipation efficiency, and solving problems such as localized overheating in traditional straight channels or serpentine flow channels.

The invention discloses an event flow action identification method based on a selection state space. The method comprises the following steps: step 1, carrying out noise reduction processing on original event stream data through a DBSCAN noise reduction algorithm; 2, scanning and serializing the event stream data after noise reduction through a Hilbert curve and a Zigzag curve, and converting an unstructured event stream into a serialized event stream; 3, encoding the serialized event stream data through a PointNet network to generate a spatial feature Token, extracting a time feature Token through a time aggregation module, and splicing the spatial feature Token and the time feature Token to obtain a complete spatial-temporal feature; and 4, inputting the spatial-temporal characteristics into a Bi-SSM module, and capturing bidirectional time dependence in an event stream through a bidirectional selection state space model, thereby realizing accurate recognition of the action. The method effectively solves the problems that in the prior art, calculation complexity is high, time information of event streams cannot be fully utilized, and model efficiency is low, and has the advantages of being low in calculation complexity and long in time sequence dependence modeling.

This application provides a method and apparatus for classifying encrypted traffic based on path signatures, relating to the field of artificial intelligence and also applicable to the financial field. The method includes: segmenting acquired encrypted traffic at the session granularity to obtain traffic paths; performing traffic path signatures on the traffic paths to obtain corresponding traffic path signature feature vectors; and inputting the traffic path signature feature vectors, after applying a Hilbert curve transform, into a pre-trained encrypted traffic classification model to obtain the encrypted traffic classification corresponding to the encrypted traffic. This application can generate signature information for encrypted traffic, thereby reflecting the geometric and statistical characteristics of encrypted traffic and exhibiting strong generalization ability.

The invention relates to the technical field of location service privacy protection, in particular to a location privacy protection method based on regional confusion and differential privacy, which comprises the following steps: S1, dividing regions, processing a location set uploaded by a user, constructing a probability distribution matrix, and generating a user pseudo location; and S2, request forwarding: the relay server receives a user query request and generates a hidden area, forwards the query request to the LBS server based on the hidden area, and finally returns a query result to the user. According to the method, the positions are mapped and sorted through the Hilbert curve, so that the spatial proximity of the positions is accurately reflected; and the optimal clustering number k is adaptively determined in combination with a QK-means clustering algorithm, so that PLS division is more reasonable, and the accuracy and matching efficiency of region division are improved.

The application discloses a land ownership space information construction method based on cadastral survey data, relates to the technical field of data processing, and comprises the following steps: adopting cadastral elements and ownership identification in cadastral survey results to perform coordinate unification, geometry repair, reference relationship calibration and ownership relationship calibration, and generating a cadastral space object sequence; performing Hilbert curve coding sorting on parcel space objects in the cadastral space object sequence, and dividing data sections according to cadastral sub-areas and continuous coding intervals, to generate a cadastral segmented directory; constructing a Hilbert global directory, an R-star tree local space index and a B-plus tree ownership attribute index through the cadastral segmented directory, and generating an ownership space construction index table. Through vectorized R-tree space connection, the application can perform batch judgment on the spatial relationship among parcel surfaces, boundary lines and boundary points, and can generate segmented ownership space results in combination with ownership attribute matching, so that the spatial relationship calculation efficiency and the ownership corresponding accuracy in a big data processing scene can be improved.

The application provides a kind of laser radar point cloud fast query matching method based on local sensitive hash, first through laser radar scanning collection environment data, point cloud segmentation is carried out to the point cloud data collected using K-means clustering segmentation algorithm;Then the feature of data object is extracted, and the data set S after preprocessing is obtained;The point cloud data in voxel is generated one-dimensional index sequence using pseudo-hilbert curve generation algorithm;The pseudo-hilbert curve sequence obtained is encoded using LSH, and the index is constructed, the Euclidean distance between query point cloud and candidate point cloud is confirmed, and the nearest point cloud in voxel is confirmed.The method of the application improves the speed of point cloud feature extraction query based on local sensitive hash index method, shortens the query matching time and improves the query matching precision;LSH index gathers similar point clouds together by hashing coding pseudo-hilbert sequence, reduces the number of point clouds that need to calculate distance, thereby improves the retrieval efficiency.

The invention discloses a sea wave disaster early warning information precision verification method, and relates to the technical field of intelligent management, and the method comprises the following steps: dividing a three-dimensional space-time grid, calculating sea wave energy characteristics, reconstructing the monitoring data of a problem node, generating an analysis link, and carrying out the first verification of alarm information based on the time characteristics of a sea wave disaster; and updating the alarm information according to the first verification result. According to the method, the global sea area is divided into the three-dimensional grids with the unified temporal-spatial resolution, spatial-temporal alignment of multi-source monitoring data is achieved through three-dimensional Hilbert curve coding, the precision and consistency of sea wave energy feature extraction are improved, a comparison mechanism of theoretical energy and actual energy is constructed based on an inter-node energy propagation model, and the real energy feature extraction accuracy is improved. The reliability and robustness of monitoring data are improved, the accuracy and authority of early warning information are improved by constructing an analysis link, and the node with the highest error risk is preferentially processed by adopting a strategy of firstly verifying problem nodes and then rechecking reliable nodes.

The invention relates to the technical field of geographic information processing, in particular to a real estate surveying and mapping resource sharing system and method, and the system comprises a heterogeneous primitive analysis module which constructs a spatial topological incidence matrix and generates a standardized semantic feature set, and a distributed index module which generates spatial position hash codes by using a Hilbert curve to perform fragmentation storage, the vision field verification module uses a ray crossover algorithm to calculate a space inclusion state and generate a dynamic access authorization token, and the increment synchronization module calculates a geometric element difference package to complete instant redrawing. According to the method, topological dislocation caused by source data format difference is eliminated by constructing a heterogeneous semantic mapping matrix, bandwidth occupation and loading delay are reduced based on a differential transmission mechanism of viewpoint coordinates, and self-adaptive adjustment of access granularity is realized by utilizing a dynamic geofencing algorithm; the problem of I / O blocking under high-concurrency requests is solved by adopting distributed hash fragment index logic, and accurate distribution and real-time synchronization of resources are ensured.

The invention discloses a wind power generation power estimation method based on data fusion processing, and relates to the technical field of wind power generation power prediction, and the method comprises the steps: collecting real-time multi-modal data, carrying out the preprocessing, generating a real-time feature set, extracting a power sequence, calculating the number of IMFs through a cyclic autocorrelation function and a similarity coefficient, and carrying out the prediction of the IMFs. The method comprises the following steps: performing EEMD decomposition in combination with noise amplitude to obtain an IMF sequence, dividing a wind power plant into equal-proportion grids, mapping geographic coordinates of a fan into a two-dimensional grid index, converting the two-dimensional grid index into one-dimensional codes through a Hilbert curve algorithm, constructing space-time grids in combination with time binary codes, mapping multi-dimensional feature vectors to generate an aggregation space-time feature cube, and performing power prediction; according to the method, the input quality and precision of the prediction model are improved through time alignment and feature extraction of the multi-modal data, and the prediction robustness and the system adaptability are improved by using EEMD decomposition and space-time grid modeling.

The invention discloses a large-scale spiking neural network efficient partitioning method based on a connection mode index, and the method comprises the following steps: aiming at the characteristics of brain-like computing hardware, carrying out the topological sorting, carrying out the locality-keeping index sorting of neurons through employing a Hilbert curve, and carrying out the resource-limited segmentation algorithm based on binary search, neurons with common pre-synaptic inputs are clustered into the same computing core. According to the method, a pulse sharing mechanism of hardware can be utilized to the maximum extent, the pulse broadcast flow of the network-on-chip is remarkably reduced, and core resource waste caused by axon memory limitation is reduced. Compared with the prior art, the compiling speed is increased by several orders of magnitude when the network with the gigons of neuron scale is processed, and the communication delay and the energy consumption during operation are greatly optimized at the same time.

The invention provides an anti-counterfeiting image texture feature extraction method and system, and the method comprises the steps: carrying out the YCbCr color space conversion of an anti-counterfeiting image, carrying out the two-dimensional dual-tree complex wavelet transformation of Y, Cb and Cr components, and obtaining a multi-scale multi-direction high-frequency sub-band coefficient matrix; energy of each high-frequency sub-band is calculated, global texture complexity parameters are determined based on the energy, a dynamic selection threshold value is generated, and the high-frequency sub-bands with the energy exceeding the threshold value are effective texture sub-bands; for the effective texture sub-band, calculating information entropy, determining the size of a sub-block, dividing the sub-block, and reserving the sub-block of which the standard deviation is greater than the mean value of the standard deviation of the sub-block in the sub-band; the effective texture sub-band reserved sub-block coefficients are scanned and spliced into a one-dimensional feature vector according to a Hilbert curve, sliding windows with odd lengths are used for traversing, a bit 1 is generated when window center coefficients are larger than a left neighborhood coefficient mean value and a right neighborhood coefficient mean value at the same time, otherwise, a bit 0 is generated, and an anti-fake image texture feature hash sequence is generated through splicing.

The present application provides a gas detection method and a gas detection device, the method comprising acquiring gas spectral data, wherein the gas spectral data is obtained by detecting a mixed gas in a gas pool using a quantum cascade laser, and the mixed gas includes a gas to be detected; obtaining a plurality of local index information based on global index information of the gas spectral data, the global index information representing coordinate curves of a plurality of signals in the gas spectral data in different coordinate system quadrants; performing quadrant flipping and rotation processing on a coordinate system quadrant of at least one local index information based on a Hilbert space mapping method to obtain an initial global coordinate; using the initial global coordinate as new global index information to calculate a new initial global coordinate, and determining the new initial global coordinate corresponding to a target order of the Hilbert curve as target spectral data; inputting the target spectral data into a gas concentration detection model, and outputting gas concentration information of the gas to be detected.

The invention provides a three-dimensional geological block bidirectional mapping method based on a W-Hilbert curve, and relates to the field of geological data coding, and the method comprises the steps: obtaining a to-be-coded three-dimensional geological block, and carrying out the W-Hilbert coding, and obtaining a to-be-converted code; constructing a positive mapping algorithm; mapping the to-be-converted code to a storage position through a positive mapping algorithm; constructing an inverse mapping algorithm; and the storage position is converted into the W-Hilbert code through an inverse mapping algorithm, so that seamless conversion between the W-Hilbert code and the storage position is realized. And the conversion between the W-Hilbert coding and the storage index is realized by respectively calculating the coding position relationship at each level.

A file is received that includes locus information. The received file is preprocessed, including by selecting a size for a Hilbert curve. File contents are transformed using the Hilbert curve. The transformed file contents are down-sampled. A hashing operation is performed on the down-sampled result.

The invention provides a flow channel structure and a liquid-cooling cold plate, the flow channel structure comprises a fractal curve flow channel, the fractal curve flow channel is a Hilbert curve flow channel or a Petronol curve flow channel, and the fractal curve flow channel is used for increasing the heat dissipation area and strengthening the convection heat exchange efficiency in a limited plane space; wherein the Hilbert curve flow channel is constructed on the basis of a Hilbert curve, and the Peano curve flow channel is constructed on the basis of a Peano curve. According to the flow channel structure and the liquid-cooling cold plate provided by the embodiment of the invention, the heat dissipation performance is improved.

This invention discloses a data completion method based on spatiotemporal privacy protection, which relates to the field of data completion technology. The method comprises the following steps: obtaining spatiotemporal data to be completed; obfuscating the spatiotemporal data to be completed; mapping the obfuscated spatiotemporal data onto a three-dimensional Hilbert curve, assigning a unique Hilbert value to each position, and constructing a location privacy set; generating a confusion probability matrix by permuting and flipping the location privacy set; and inputting the confusion probability matrix, the obfuscated spatiotemporal data, and the spatiotemporal data to be completed into a completion model to obtain the completed spatiotemporal data. The completion model includes a diffusion process and an inverse process. The present invention improves completion accuracy in privacy protection scenarios.

The invention provides an event flow action identification method and system based on a frame-point synergistic effect, and relates to the technical field of computer vision. The method comprises the following steps: extracting context event points from an original event stream by using a pulse type context learning device, and adaptively selecting a plurality of key event points from the context event points; sorting the key event points according to a Hilbert curve to reserve a spatial topological structure, and modeling long space-time dependence of the key event points to obtain event point features; converting the original event stream into a structured event frame through a time accumulation window; performing feature extraction on the event frame based on a Transform network to obtain an event frame feature; and fusing event point features and event frame features, introducing a CLIP text encoder to generate frame / point related semantic prompts, and finishing action recognition by comparing the consistency of loss constraint event features and text semantics.

The invention relates to a three-dimensional point cloud semantic segmentation method based on complementary ordered modeling and dynamic space coding, and belongs to the field of computer vision and graphic images. The method comprises the following steps: firstly, carrying out complementary ordered modeling on input three-dimensional point clouds in a three-dimensional point cloud data set by fusing a Z-order curve, a Hilbert curve and a mirror image variant thereof, and generating multi-serialization point clouds with complementary structure retentivity; the spatial coding dimension is adaptively adjusted through dynamic spatial coding, and spatial enhancement information is obtained; establishing a spatial perception aggregation module including local feature propagation and global context perception, and performing point cloud global-local feature aggregation on the spatial enhancement information to obtain a three-dimensional point cloud fusion feature; and finally, performing three-dimensional point cloud segmentation on the three-dimensional point cloud fusion feature to obtain a point cloud semantic tag. The method can effectively improve the precision of point cloud semantic segmentation, and is suitable for three-dimensional point cloud semantic segmentation tasks in scenes of intelligent buildings, automatic driving, augmented reality and the like.

This invention provides a cartographic method for optimizing real-time map rendering, belonging to the field of cartographic technology for real-time map rendering optimization. This invention pre-calculates the screen space projection area at the central processing unit (CPU) and repackages it into batches with varying loads, then asynchronously submits the batches to the graphics processor (GPU) via a drawing indirect instruction mechanism. It uses 3D Hilbert curves to map tile coordinates as linear indices, combining quadtree divide-and-conquer recursion with user motion prediction to drive tile prefetching. It reconstructs low-resolution tile textures using a bi-branch map texture super-resolution fusion model and adaptively adjusts the high-frequency gating weight inference scaling factor through a comprehensive quality adjustment value. It compresses the occlusion query overhead per frame through a bi-phase occlusion culling pipeline combined with temporal reprojection. This invention solves the technical problems of unbalanced geometric subdivision load, low tile scheduling cache hit rate, and difficulty in adaptively balancing texture super-resolution reconstruction quality and inference latency during real-time map rendering.

The application relates to a spatial range query method based on HQuadTree and belongs to the technical field of spatial database index. The application comprises two parts of spatial index construction and spatial range query. In the index construction, Hilbert curve and QuadTree are combined to create HQuadTree index. In the query execution, a query q is given, a deep-first combined limited-breadth expansion strategy is adopted to moderately drill down the spatial range query, and the number of invalid objects returned can be effectively reduced. In addition, in order to avoid the intersection judgment times in the drilling process, a fast intersection judgment strategy is proposed to reduce the judgment times, so that the query efficiency can be effectively improved.

The invention discloses a new energy output scene generation method based on meteorological factor guidance, and relates to the technical field of scene generation, and the method comprises the following steps: S1, constructing a pre-training denoising diffusion probability model based on Hilbert curve projection distance and autocorrelation coefficient feature guidance, and achieving the rolling generation of continuous multi-day weather scenes; s2, freezing the pre-training de-noising diffusion probability model, adding a ControlNet control branch, inputting the weather sample with the condition label into the ControlNet control branch, and training to obtain a weather scene generation model; and S3, a weather-to-power conversion model is adopted, the generated weather scene is converted into new energy output, and a continuous multi-day new energy output scene meeting the condition control requirement is obtained. According to the method, probability distribution and correlation targets are considered, and a diversified scene which has high probability distribution consistency and time sequence correlation with a real scene in the next day can be generated.

This invention relates to the field of image encryption technology and addresses the problems of slow encryption speed and weak resistance to attacks in existing image encryption methods. The image encryption method described in this invention uses a ternary fractional discrete chaotic neural network system to generate a chaotic sequence related to the plaintext. A fractal sorting and scrambling process is then performed on the original image using a fractal model approach. To achieve better results, a double scrambling process is applied to both rows and columns. Finally, a synchronous scrambling and diffusion operation is performed according to the Hilbert curve traversal order, which can simultaneously change the position and size of pixel values. This method is the first to apply fractal thinking to the scrambling process in encryption, achieving good scrambling results. The synchronous scrambling and diffusion operation further improves the encryption efficiency to a certain extent.

The invention provides a multi-scale geological block cross-scale indexing method, a medium, equipment and a product, and relates to the technical field of spatial indexing.The method comprises the steps that geological blocks of different scales are mapped into a unified data structure, the blocks of the same spatial area under different levels are continuously coded along a W-Hilbert curve, and the blocks of the same spatial area under different levels are obtained; generating globally unique codes of the parent block codes and the sub-block codes; the multi-scale block is stored according to W-Hilbert coding, a sub-block and father block query algorithm is constructed based on coding offset through a hierarchical offset relationship of coding, and a mapping relationship between father block coding and sub-block coding between blocks of different scales is established. According to the method, efficient cross-scale indexing of the multi-scale geological block can be realized.

The invention belongs to the technical field of optical fiber communication, and particularly relates to a high-reliability multi-dimensional encryption method and system. The multi-dimensional encryption method comprises the following steps: generating a high-reliability chaotic key through a random feedback dynamic cascade chaotic system; according to the generated high-reliability chaotic key, bit dimension encryption is carried out through a DNA encryption method integrating base mapping and base calculation; symbol dimension encryption is realized through lossless low-complexity constellation rotation; and random parting scrambling is carried out on the data through a high-reliability chaotic key and a Hilbert curve. According to the optical communication system-oriented signal transmission encryption method and system, a random feedback dynamic cascade chaotic key generation system is designed, bit and symbol level encryption is performed on a signal through a generated chaotic key, then a traversal matrix is generated according to a Hilbert fractal rule, random scrambling of a signal sequence is further realized, and the random scrambling of the signal sequence is realized. The method can provide safe and reliable encryption transmission for data, and has great potential and application prospects in the field of optical fiber communication.

The application provides a point cloud semantic segmentation method based on a state space model, and relates to the technical field of point cloud semantic segmentation. The application aims to solve the problems of long-range information collapse, single serialization bias and insufficient local feature aggregation capability of the state space model in large-scale point cloud sequence modeling. The application stacks a sub-manifold convolution layer after the state space module, uses the convolution locality bias to relieve the long-range information attenuation, adopts a staggered Hilbert serialization strategy, randomly switches the Hilbert curve of three axis sequences to reduce the spatial bias and enhance the spatial diversity, and designs an adaptive down-sampling module, uses the selective state mechanism to aggregate the features and retains the tail points as the output, so as to comprehensively improve the global and local modeling capability. The method has the advantages of fast reasoning speed and low memory occupation while ensuring excellent segmentation accuracy, and is suitable for large-scale complex scenes.