404 results about "Octree" patented technology

Provided are methods and apparatuses for encoding and decoding three-dimensional object data, which consists of point texture data, voxel data, or octree data. The method of encoding three-dimensional object data involves generating three-dimensional object data having a tree structure in which nodes are attached labels indicating their types; encoding nodes of the three-dimensional object data; and generating the three-dimensional object data whose nodes are encoded into a bitstream. The apparatus for encoding three-dimensional object data includes a tree structure generator which generates three-dimensional object data having a tree structure in which nodes are attached labels indicating their types; a merging order selector which merges the nodes of the three-dimensional object data by referring to their labels; a node encoder which encodes merged nodes; and a bitstream generator which generates the three-dimensional object data whose merged nodes are encoded into a bitstream. The method of decoding three-dimensional object data involves reading continue flag information from a bitstream of encoded three-dimensional object data and decoding the continue flag information; decoding note type information of the bitstream; decoding an 'S' node if the note type information indicates that a current node is an 'S' node and decoding a PPM node if the note type information indicates that the current node is a PPM node; and restoring the three-dimensional object data whose nodes are encoded to a tree structure. The apparatus for decoding three-dimensional object data includes a bitstream reader which receives a bitstream of encoded three-dimensional object data; a node decoder which decodes the bitstream; and a tree structure restorer which restores decoded nodes to a tree structure.

The invention discloses a multiline laser radar-based 3D point cloud segmentation method. The method comprises the following steps of: 1) scanning 3D point cloud data in a 360-degree range by utilizing multiline laser radar, establishing a Cartesian coordinate system OXYZ, converting the 3D point cloud data under the Cartesian coordinate system, and pre-processing the 3D point cloud data under the Cartesian coordinate system so as to determine a region of interest in the 3D point cloud data; 2) filtering suspended obstacle points in the region of interest by utilizing the statistic characteristics of adjacent points; 3) constructing a polar coordinates grid map, mapping the 3D point cloud data, the suspended obstacle points of which are filtered, into the polar coordinates grid map, and segmenting non-ground point cloud data from the 3D point cloud data in the polar coordinates grid map; and 4) voxelizing the non-ground point cloud data by utilizing an octree and carrying out clustering segmentation by adoption of an octree voxel grid-based region growing method. The method can improve the operation efficiency, is high in detection precision and strong in reliability, and can be widely applied to the technical field of vehicle environment perception.

A method of propagating a ray through an image includes providing a digitized volumetric image comprising a plurality of intensities corresponding to a domain of voxels in a 3-dimensional space, providing a reduced path octree structure of said volumetric image, said reduced path octree comprising a plurality of first level nodes, wherein each first level node contains a plurality of said voxels, initializing a ray inside said volumetric image, and visiting each first level node along said ray, wherein if said first level node is non-empty, visiting each voxel contained within each first level node.

The present invention provides a browser based mass three-dimensional point cloud data release method. The method comprises the following steps that S1. a bounding box of original three-dimensional point cloud data is calculated, and block division is carried out on the original three-dimensional point cloud data according to the bounding box and a regular grid division theory, so as to divide a plurality of block point cloud bounding boxes; S2. an octree is established for each block point cloud bounding box, and a description file is configured for each octree; S3. regular grid division is carried out on a point cloud octree, and a quadtree structure is established for the octree contained in each regular grid; and S4. a browser sends in real time a query request to a server, the server returns in real time an octree query result to the browser, the browser carries out visibility determination on the octree query result, and the visible octree is loaded, so as to complete dynamic loading and scheduling of mass three-dimensional point cloud data. According to the method disclosed by the present invention, web based release and visualization of mass large point cloud files are realized, and the problem that point cloud data is difficult to release is solved.

The embodiments of the invention provide a spatial index construction method and device for mass point cloud data. The method comprises the steps of performing partitioning treatment on acquired original point cloud data to obtain a plurality of point cloud data blocks; for each point cloud data block, establishing an octree index of the current point cloud data block; performing combining treatment on the octree indexes of the multiple point cloud data blocks to obtain a spatial index structure of the original point cloud data. According to the technical solution, the conventional octree index structure is improved; partitioning treatment is performed on original point cloud data, so that mass point cloud data can be disassembled and thus memory space occupation in spatial index construction is reduced; after multiple octree indexes are combined, an octree index where target data are located can be found first and then target data can be searched for based on the found octree index in later query, so that the query efficiency is greatly increased.

A method for optimizing the development of a fluid reservoir using a fractured medium mesh generated from a first-order balanced octree technique is disclosed. A mesh of a discrete fracture network is generated by defining a set of cells for each fracture. A mesh of the matrix medium is then generated by dividing each cell by an octree technique, wherein a cell is divided into eight cells. The cells resulting from the division are themselves split in eight, until each cell resulting from a division comprises a number of fractures below or equal to a given threshold. Transmissivities between the cells of the fracture mesh, transmissivities between the cells of the matrix medium mesh, and transmissivities between cells of the fracture mesh and cells of the matrix medium mesh are then determined. Finally, the cells and the transmissivities are used for generating an image of the fluid reservoir from which the development of the fluid reservoir is optimized.

A family of node structures for representing 3-dimensional objects using depth image are provided. These node structures can be adopted into MPEG-4 AFX for conventional polygonal 3D representations. Main formats of the family are DepthImage, PointTexture and OctreeImage. DepthImage represents an object by a union of its reference images and corresponding depth maps. PointTexture represents the object as a set of colored points parameterized by projection onto a regular 2D grid. OctreeImage converts the same data into hierarchical octree-structured voxel model, set of compact reference images and a tree of voxel-image correspondence indices. DepthImage and OctreeImage have animated versions, where reference images are replaced by videostreams. DIBR formats are very convenient for 3D model construction from 3D range-scanning and multiple source video data. MPEG-4 framework allows construction of a wide variety of representations from the main DIBR formats, providing flexible tools for effective work with 3D models. Compression of the DIBR formats is achieved by application of image (video) compression techniques to depth maps and reference images (videostreams).

The invention discloses a scattered point cloud compression method based on an octree. The method aims at three-dimensional point cloud data without any precognition information. According to the method provided by the invention, an octree segmentation stopping condition is improved, the segmentation can be stopped at a suitable depth and sizes of voxels are ensured to be suitable. A k-neighborhood is established on the basis of segmentation. Outliers of an original point cloud are removed through utilization of a simple and effective statistic method. According to a data structure, bit masks are allocated to each node. Data is searched and operated by controlling the bit masks when traversal is carried out, and follow-up point position coding is optimized. According to the method, the outliers and surface miscellaneous points are effectively removed, and the point cloud compression efficiency at interval coding is improved at interval coding.

The invention discloses a three-dimensional space index method aiming at massive laser radar point cloud models. Aiming at characteristics of massive data and high spatial resolution of point cloud models, the method indexes data to a single three-dimensional space point by adopting a multi-level hybrid index strategy; aiming at an integrity characteristic of point cloud models, the method continuously subdivides massive cloud points from a minimum bounding rectangle (MBB) of an overall point cloud model by adopting a cotree index, and three-dimensional points in cotree index nodes are distributed uniformly in space; and aiming at a scattered characteristic and the data post-processing requirement of large-scale point cloud models, the method indexes a single three-dimensional space point by adopting a three-dimensional K-D tree, and the quick inquiry and processing of a single point coordinate and the attribute data of the single point coordinate can be achieved. On the basis of above integrated space index construction, the method adopts a three-dimensional R tree to manage the MBBs of a plurality of point cloud models in three-dimensional scenes, and a multi-level hybrid space index mode is formed finally.

The invention discloses a vehicle-mounted road scene point cloud automatic classification method and system based on deep learning, and the method comprises the steps: building a training sample set:collecting the three-dimensional point cloud data of a road scene vehicle-mounted laser, labeling a category label, and packaging the three-dimensional point cloud data and corresponding category label information as a point cloud data set; Designing a three-dimensional deep learning network model based on an octree data structure, wherein the three-dimensional deep learning network model adopts aU-shaped full convolutional neural network model; A loss function for weighting each category is utilized to solve the problem that the number difference of different categories of points of the vehicle-mounted laser point cloud of the road scene is large, a network model is trained based on a training sample set, and a trained point cloud classification model is obtained; And inputting to-be-classified vehicle-mounted laser three-dimensional point cloud data. By adopting the technical scheme of the invention, the laser point cloud under the vehicle-mounted road scene with different types, numbers and sizes can be automatically classified stably and accurately.

A method for the problem of reconstructing a watertight surface defined by an implicit equation from a finite set of oriented points. As in other surface reconstruction approaches disctretizations of this continuous formulation reduce to the solution of sparse least-squares problems. Rather than forcing the implicit function to approximate the indicator function of the volume bounded by the surface, in the present formulation the implicit function is a smooth approximation of the signed distance function to the surface. Then solution thus introduced is a very simple hybrid FE/FD discretization, which together with an octree partitioning of space, and the Dual Marching Cubes algorithm produces accurate and adaptive meshes.

The invention discloses a scanning laser based large-scale three-dimensional terrain modeling method. The scanning laser based large-scale three-dimensional terrain modeling method comprises the following steps of: firstly perceiving a surrounding environment through the scanning laser arranged on a mobile robot platform, and transforming original laser data into three-dimensional point clouds through coordinate transformation; secondly preprocessing the three-dimensional point clouds by adopting a median filter method, and constructing an octree structure to complete data storage; then carrying out fusion processing on data by adopting a Gaussian mixed model based randomization method to build a terrain environment model in real time; and finally detecting and processing an obstructed area positioned in the terrain environment model to further enhance model accuracy. In the invention, the problem of real-time modeling of a large-scale complex terrain is effectively solved by adopting the low-cost two-dimensional scanning laser; and good balance between the model accuracy and real-time property is realized in such a way that randomization processing is carried out on various uncertainties of the terrain environment model, and therefore, wide requirements of different fields, such as a GIS (Geographic Information System) system or unmanned autonomous navigation and the like, can be met.

The invention discloses a three-dimensional point cloud reconstruction method based on improved shield poisson algorithm. The method comprises the following steps: obtaining a three-dimensional point cloud data of a model; introducing shielding factor, and preprocessing the constraint of the poisson algorithm by the constraint of the introduction point and the gradient; using the octree segmentation to simplify the denoising of the point cloud, compressing the storage, establishing the inter-cloud topology, and realizing the fast establishment and efficient searching of the octagonal tree neighbor nodes; normalizing each point cloud vector to reduce the ambiguity of normal direction; calculating the vector field; solving poisson equation to get instruction function; on point cloud equivalent surface extraction, getting three-dimensional reconstruction model. The method of the three-dimensional point cloud reconstruction method based on improved shield poisson algorithm has the advantages of improving the normal vector precision, removing the pseudo-closed surface well, and being well filled with the hole, and the detail of the object is rich. The method can be widely used in the fields of reverse engineering, cultural restoration, and medical image.

A method for constructing a lookup table for converting data from a first color space to a second color space includes generating first data in the first color space corresponding to an input color gamut; representing the first data in a first non-uniform three-dimensional octree structure, wherein an amount of data at any leaf node does not exceed a predetermined threshold; converting the first data to second data in the second color space; and generating a non-uniform three-dimensional octree lookup table relating the first data represented by the first non-uniform three-dimensional octree structure to the second data.

The invention relates to a color attribute encoding method based on a TMC3 point cloud encoder. According to the TMC3 flow, the invention encodes the position information by using an octree structurewhile reserving the encoding sequence. After that, the centroid of the position information is calculated and kd-tree is used to find the point closest to the centroid as the first point in the detaillayer (LoD) generation process. The approximate color value and the predicted color information are differentially encoded. This method reduces the coding rate of color information by using the feature that the difference between the approximate color value and the predicted color value is reduced from the center of mass. The experimental results show that the invention achieves better effect than TMC3 in chromaticity information processing.

The present invention relates to a three-dimensional model progression transmission method on basis of keeping character, the present invention holds the improvement for character keeping on the basis of the simplified arithmetic of the original three-dimensional model, and at the same time of simplifying the geometrical model, the present invention keeps the topological property and attribute character of the model, and gets the underlying net with smaller data amount, and on the basis of above, the processing coding is held distributed on basis of octree to construct the procession net document. When in transmission, the octree broadness prior coding underlying net is firstly used, and then adopting the procession net transmission method to transport a serial detail recovering information, and continuousely improving model presicion and recovering the original document, thereby solving the problem that: the responding time for the traditional transmission method is long. In adition, the viewpoint related transmission strategy facing the object is designed for the three-dimensional scene, increasing the actural drawing, and making the visibility judging of the object is carried out by putting on the clinet end, and reducing the loading of the server.

The invention relates to a label collision prevention technology in radio frequency identification and particularly relates to a multi-way tree anti-collision algorithm applicable to an RFID system. The algorithm comprises the steps of appointing the algorithm, initializing prefix stacks, detecting collisions, selecting search modes, determining prefixes, judging prefix stacks, enabling prefix stacks to be null finally, and ending the algorithm. According to the algorithm, a reader-writer is used for judging numbers of adjacent collision bits, an octree search mode, a quadtree search mode or a binary tree search mode free of free timeslots is selected adaptively, the search speed is accelerated, transmission of subsequent data is cut off immediately when three adjacent collision bits are judged, and a counter is used for feeding back the lowest bits in adjacent collision bits within transmission time of a plurality of bits, so that the transmission of the data is reduced, the energy consumption is reduced, free time slots are eliminated, and the identification time is shortened.

An encoder is configured to compress spatial information for points included in a three-dimensional (3D) volumetric content representation using an octree, predictive tree, or other geometric compression technique. For points of the 3D volumetric content that are spatially located as same or similar locations in 3D space, such duplicated points, may be signaled using a duplicate point count. The duplicate point count may be used instead of explicitly signaling (duplicated) spatial information in the predictive tree for the duplicated points, as an example. Similarly a decoder is configured to generate a reconstructed three-dimensional representation of the volumetric content that includes the duplicate points, wherein the reconstructed 3D representation is generated from a bit stream including one or more duplicate point counts.

An encoder is configured to compress point cloud geometry information using an octree/predictive tree combination geometric compression technique that embeds predictive trees in leaf nodes of an octree instead of encoding additional octree occupancy symbols for the leaf nodes. Alternatively an encoder may be configured to embed octrees in leaf nodes of a predictive tree structure. Similarly a decoder is configured to generate a reconstructed three-dimensional geometry from a bit stream including combined octree and predictive tree encoding information.