GIS technology based 3D vector map rendering method of dynamic LOD

This paper analyzes the factors simplify the map elements from the map operation and visibility of the start, and the corresponding cache structure designed based on the results of the analysis to speed map rendering. Efficient LOD of trapezoidal grid environment under study GPU, recombinant Discussion simplify updating before and after the memory node and the data EBO final vector map made based on a dynamic three-dimensional rendering method LOD GIS technology.
Factors 1 simplified
1.1 operational impact on the simplified map
proposed method based on a simplified view, in real time to simplify the operation when the position of the elements of the viewpoint changes. When users map operation, performed when the distance between the viewpoint and factor changes simplify, you can use the cached data is drawn when the change does not occur, thus speeding up the rendering. Therefore, changes need to be discussed from the point of view of the elements of map operations caused.
(1) shift operations
shift operation may result from the point of view of the elements of change. In a non-pitching state, from the viewpoint shift operation does not cause a change to the feature; in pitch state, the translational distance between the viewpoint and the operating element can cause changes.

Schematic panning operation
(2) rotating
the rotation action may cause the viewpoint and the distance element is changed. The non-pitch state, the rotating operation and the element does not result from a viewpoint is changed; the pitching state, the rotation action causes the element from the viewpoint changes.

A schematic view of the rotational operation
(3) zooming operation
zoom operation causes the distance between the viewpoint and the element is changed.

Scaling a schematic view of the operation
(4) the pitch
the pitch distance between the viewpoint and the operating element can cause changes.

Schematic tilting operation
The foregoing analysis shows that in a non-pitching state, map pan operation and rotational operation element does not change the distance from the viewpoint, can be used on a map cache a draw, which accelerate the rendering speed; for low pitch angle, pan, and little effect on the rotating operation is simplified, can be used on a drawing data. And for tilt, and zoom operations, elements severely affected right viewpoint, it is necessary to simplify the operation, as shown in the following table.

1.2 Effect elements visibility simplified
static LOD does not require real-time map simplification, the cached data is directly drawn simplified, extremely small feature dimensions such that its apex is extremely simplified with less data, and therefore the elements which seldom be drawn on the impact of static method LOD rendering efficiency is not obvious. The dynamic simplified approach requires all the elements within the viewport simplify, simplify the more efficient operation is extremely sensitive to the number of features to simplify the participation of elements of the simplified longer, when an element is extremely small can lead to unnecessary streamline operations resulting in rendering efficiency low. Therefore, to ensure drawn on the basis of the results should be involved as much as possible to reduce the simplified dynamic elements.
The scale of change will lead to the presentation of geometric elements changes state, when the scale is extremely small portion of the human eye can not recognize geometric elements on the screen, these features are drawn or not drawn will not have much impact on the results. Based rendering strategy traditional LOD method, the elements as long as it needs to be presented within the viewport, paper elements presented state setting visibility threshold value, when the pixel size of the element rendered smaller than the threshold value is no longer drawn, thereby reducing the dynamic simplified participation elements of data, improve the efficiency of map drawing.

Visibility threshold associated mapping results
2 based vector map grid trapezoidal method LOD
LOD 2.1 trapezoid grid
In this paper, the basis sequence table storage trapezoidal split sequence ladder, according to the order of the node weight using the original node sequence list is stored. In the ladder of simplified process, the cursor stored list of the current reduced state. If the next operation after the map, the list is less than the tolerance of simplifying the current cursor tolerance, then the query list up, simplify operation; if the next operation map, the tolerance is larger than the list simplify the current cursor tolerance, then the query list down for restore operation. By simplifying the elements of the state (ie, the list cursor) of storage, avoiding thereby speeding up the global search simplified speed. As shown below, the current cursor is Pcursor, located in the node weight chain + of Pn 1, if Pcursor less than Pn + 1 tolerance, the search direction based on the current trapezoid grid simplify operation; if Pcursor not less than Pn + 1 tolerance, down through a trapezoidal restore operation based on the current grid.

2.2 EBO GPU data updating method based on
simplification algorithm based on the trapezoidal grid understood, the number of node locations trapezoidal or trapezoidal ladder network will change after the simplified pattern. Firstly, the search range within the object space by the index, according to a simplified algorithm trapezoidal grid of grid trapezoidal simplified. By traversing trapezoidal trapezoidal grid simplified, obtaining a trapezoidal index number and the status of the node, from top to bottom recombinant trapezoidal grid and the eventual update of the data memory in EBO. Hereinafter, this paper describes the program from the cache data updating method and EBO trapezoidal grid.
(1) Design apex storage
GPU vertex buffer objects used VBO (Vertex Buffer Object) stores vertex data, using the index stored in the index data cache object EBO. This design of single and multiple caches object storage cache object store two storage scheme used to implement the data storage node and the node of index data. Multi-cache object stores a program created for each element of a vector VBO and a geometric EBO, single-layer cache object stores the program data stored in one node VBO, a layer of index data stored in one of the EBO.
Multiple object storage cache scheme single element index data stored in a node EBO object, the index number is not affected by elements other elements, numbered from 0. Multiple caches object storage, when certain elements are not easy to use a simplified cache object, the drawing process having a lot of cache object binding, unbound and drawing operations, resulting in an overall rendering efficiency is relatively low, which memory structure shown below .

Single cache object stores a program of all the nodes in the same layer in a unified storage VBO object, all indexes in a unified storage EBO objects, features and elements between the index needs to be offset. Increasing single cache objects stored simplified scheme of FIG local cache does not occur reusability, there is only drawn once during binding, unbound and drawing operation, the higher efficiency is plotted. However, if some elements of the layer occurs progressively,-EBO data needs to be reconstructed, which memory structure shown below.

(2) a trapezoidal index updates grid elements
when the user performs map operations, elements may be converted from one state to another. Simplification algorithm based on the trapezoidal grid understood, by simplifying trapezoidal grid between trapezoidal or trapezoidal combined, adjusted, degradation, ablation procedure is done, the operation is simplified essentially trapezoidal grid node to restructure avoid the conventional method Partition duplicate data, only the index array trapezoidal nodes are updated to complete the update EBO data transmission and does not need to create a cache object vertex data, IO memory reduces the time and improve the efficiency of the rendering.
For trapezoidal grid line, when the connection style gradual oblique angle, the trapezoidal merging directly; when in a straight angle or a rounded connecting gradual connection, consider triangle connecting straight angle or a rounded triangle fans connected generated generated index Straight connector offsets need to consider the number of fan-generated number of nodes or a rounded triangle vertices connected generated, as shown in FIG.

For trapezoidal grid surface, as shown in Figure 1: P1 to P11 of the original node, Pm to Pm + 8 split produced a trapezoidal node. FIG front as an example simplified partial trapezoid trapezoidal grid merge operation occurs, by traversing the sequence table trapezoidal trapezoidal its visibility determination, and obtains the index number of nodes to reconstruct the index trapezoidal array, a finally updated EBO. The results shown in simplified view of a simplified two data before and after the change as shown in FIG.

FIG simplified schematic side trapezoidal grid

FIG trapezoidal dihedral grid index update schematic
seen before and after the update methods described herein need not be retransmitted vertices, VBO same data, only the update data of EBO to complete trapezoidal grid data before and after update simplified, reduced drawing time consuming process memory IO stage, accelerated rendering.
LOD 3 vector map rendering method based on GPU
3.1 vector-based map drawing process LOD trapezoidal grid
for LOD conventional methods, and simplifies the separation of the split problem, we design drawing process based on the vector map LOD trapezoidal grid. Data is completed when the initial load split data, calculation of the weight, build, and vertex data transmission trapezoidal grid. In the process of rendering the map, different maps according to operation of determining the need for a map or feature to simplify, or repainted without elements for simplifying some cases do not need the map, or the map cache design elements, directly on the data of one frame drawing, thus speeding up the rendering speed. Using the trapezoidal grid simplified quick operation is required to simplify the map, this process is not performed and the vertex split transmission, and the data plotted EBO simply update to complete the rendering of the map, the map accelerate the rendering speed. As shown below.

3.2 caches and cache design elements
based on the above analysis shows, the map for the operation, when the lower case and a non-pitching pitch, the translational and rotational operation does not occur simplified map, can be used to cache data on a direct draw, improve its rendering speed. The following were designed caches LayerCache structure and feature cache Segment structure to improve map rendering speed.
(1) Segment feature cache design
feature cache stores the index data, this element comprising vertices starting position, the starting position of the index, the number of vertices and the number of indexes and index arrays. If this feature simplifies the use of the elements does not occur directly to data cache index layer cache reconstruction, by using the feature cache to speed up rendering speed, elemental structure shown below.

(2) caches Design
This design layer cache node data stored in a map layer and index data. Segment buffer layer comprises a plurality of, using direct cache when a layer is not drawn simplified, when only part of the elements needs to be simplified to simplify the elements to be updated, the cache can speed using the map rendering speed, the layer buffer structure is illustrated as follows shows.

4 Conclusion
Based on the trapezoidal grid, the vector is designed based on the dynamic map LOD GPU rendering environment. First map to explore the impact of the operation on the map by a simplified analysis of the factors affecting simplified map. Then, select the appropriate filter elements threshold value, and further describes a method for updating data before EBO and simplifying policy LOD trapezoidal grid. Finally, the proposed dynamic vector-based map trapezoidal grid LOD rendering method.
Original from: http://www.ztmapinfo.com/blog/index.php/article/37.html