A new type of mass structured data duplicate data check

In this study, the data processing method of a large-scale isoface is used to reduce the storage space of internal memory, improve data structure, and achieve predictability. The design of an algorithm mainly includes the following tasks: increasing the operation speed by improving the parallel granular of the GPU, finishing the segmentation of the adaptive tetrahedral octree, calculating the dual points through the quadric error function in four-dimensional space, searching for the minimum edge as well as finishing the data structure of the edge and the design of the octree node, and proposing that the algorithm be constantly performed until all levels of the minimum edge are found. The result of the algorithm design for the large-scale data is that the speed of parallel algorithms improves and the effect becomes more obvious. Research on large-scale data access high-speed ratio and structure improvement have experimental and theoretical reference value.


Introduction
The mode of computer storage and data organization is called data structure.Data structure refers to a set of one or more relationships between data elements.Generally, the selected data structure must lead to a higher operating or storage efficiency.The data structure is related to retrieval algorithm and index technology.The structure experience of large-scale systems verify that the degree of difficulty in achieving the quality and system configuration depends largely on the optimality of the data structure.In most cases, the algorithm is easy to obtain after the data structure is determined Occasionally, data structures are chosen in accordance with the specific algorithm.In other words, choosing the data structure that fits is crucial.In this study, a large-scale mesh simplification and an adaptive isoface-generating algorithm are investigated.Rossignac, Borrel, and Kok-Lim proposed a mesh resampling method that deals with any form of grid data.

Design of adaptive octree algorithm
Most adaptive isoface algorithms are not guaranteed to produce a mesh manifold.To solve this problem, we propose a sorting method that simplifies and reduces the error produced by the volume data function.This method can be improved by exploiting the parallel granular and operating speed dynamic increase.The splitting of the body follows a similar method.

Tetrahedral generation of adaptive octree
Each polyhedron generates a dual point.The triangle is obtained from the dividing face connected with this dual point.

Search for minimum edge
In the octree, the minimum edge cannot be divided and is significant for the formation of the tetrahedral.The storage structure of the minimum edge is similar to that of an octree.The data structure of the edge includes the direction of edge and the level and index of the octree node.To obtain it, we divide the edge into x, y, and z axes.Figure 3 illustrates the data structure of the edge.

Execution results of algorithm design
For different models, we use the adaptive isosurface and large-scale isosurface.We test the speed of the parallel algorithm, implementation results, and execution time,see table 1 and Fig 7.In the test, the simplified algorithm of the unified cache size is set to 20,000.The results indicate that for the large-scale data, the speed of the parallel algorithm is larger and its effect is more obvious.In addition, the simplified model shows that the effect of this algorithm is better on smoother surfaces.

Summary
Most adaptive isosurface algorithms cannot be guaranteed to produce a grid manifold without self-intersection.However, a sequential algorithm based on volume data simplex segmentation can solve this problem.This study is based on sequential simplex segmentation.The method of segmentation is simplified to reduce the approximate error of the volume function.Furthermore, the method is improved and is made to fully exploit the use of the parallel granular of GPU, which increases the operating speed.We propose a complete set minimum edge search method.According to the minimum edge, the nodes of the octree undergo tetrahedral segmentation, and the tetrahedral uses a marching tetrahedron to obtain the isosurface of the volume data.In practical tests, the algorithm can produce a high-quality adaptive isosurface.

MATEC Web of Conferences
01067-p.4 DeFlorianie et al. proposed that simplifying a triangle set with the same boundary by replacing it with a triangle set in an adjacent grid.Lindstrom proposed an algorithm that processes triangular patches by batches.In the current study, we propose the use of a large-scale isoface based on streaming simplified thought and saving on the global optimal queue.

FirstFigure 1 :
Figure 1: (a) adaptive quadtree in two-dimensional space and (b) divided higher-layer nodes of the octree

Figure 1
Figure 1(b) shows the splitting of the high-layer nodes of the octree.In the figure, this node becomes a polyhedron and a portion of the face becomes a polygon.The separation of the face and that of the edge both begin from this node adjacent to the lower-layer nodes.When the face is split, it produces a dual point.This dual point consists of four vectors (x, y, z, and w), where x, y, and z are the coordinates of the three-dimensional space, and w is the density value of the volume data.The dual points are connected to each edge of the face and form multiple triangles.The triangular segmentation of the outside

Figure 2 :
Figure 2: A divided triangle facing outside (white color point indicates the dual point, and dotted lines connect the octree and the dual vertex)

FigureFigure 5 Figure 6
Figure 3 Data structure of edge

Figure 6
Figure 6 shows the data structure of the face.As it is similar to the edge to be segmented, we have to know the number of the next layer segmentation.We add the results into the array of the next layer face.The next face to be segmented contains the new face expanded from the nodes of the current layer.The method of adding new edges into the face involves computing the number of new edges and offsetting it through parallel scanning.Then, we add the new edge into the next layer array to be segmented, which includes three sections.These sections are the segmented edge of the last layer, the expanded edge of the last layer nodes, and the expanded edge of the last layer face.To obtain the pseudocode of every segmented layer, we conduct the following procedure.The segmented number of the current layer edge is obtained in parallel.The edge array is segmented in parallel according to the segmented number of the minimum edge, which is zero at the front.The minimum edge is stored into the current layer minimum edge array.The non-vanishing segmented number undergoes SCAN in parallel and is offset.The number of the new parallel edges and faces is obtained.The new edges from the expanded nodes undergo SCAN and are offset in parallel.The new faces from the expanded face undergo SCAN in parallel and are offset in parallel.The number of new edges and faces from every segmented parallel face is obtained and offset.The segmentation number of the surface undergoes SCAN and is offset.The next layer array of the edge to be segmented is entered in the graphics memory.The three parts of the edges to be segmented are added into the array.The next layer array of the face to be segmented is entered in the graphics memory.The tow parts of the edges to be segmented are added into the array.At the start, we produce the edge and the face of the first layer.Through continuous and iterative execution of

Figure 7
Figure 7 Isosurface of the model: (a) isosurface of a Blunt model using a marching tetrahedron, (b) isosurface of a Blunt model using this algorithm, (c) isosurface of an engine model using a marching tetrahedron, and (d) isosurface of an engine model using this algorithm.