|
|
|
Poster Session 6
|
Optimization of Building Triangulated Irregular Network
V. Steps on non-overlapping multi-region building TIN
Because general process of building network has been discussed in another paragraph, this paragraph will give simple descriptions on non-overlapping multi-region building TIN.
-
Data pretreatment process
- Removing redundant points and reducing unnecessary computation.
- Blocking the data, making proper size of each block.
- Assigning each block a flag value to show if the block has been searched.
- Building network process
- finding the first triangle, making its longer side to be the original extending side.
- Making the original side to be the diameter of a circle, then searching within areas of the circumrectangle of the circle. After finishing searching, if a point is found that could make the angle bigger than 900, then it's the extending point to be the original side. Starting with the longer side of the new triangle, repeat step b. Or, starting with step c.
- If it could find no point or acute angle point within areas of the circumrectangle, then starting new searching areas based on conditions of the smallest angle or acute angle in building TIN.
- Searching blocks surrounding extending side. After completing an area, it computers new searching blocks based on the biggest angle condition to reduce searching range. When the two confirm to each other, it end the searching.
- Assigning all areas to no searching flag.
- Repeating step b to step e until the TIN of the whole area is built.
VI. Algorithm Optimization Analyzing and Assessing
Through the comparisons of the three ways, we can find the third way has high efficiency, relatively simple data structure, so it can be taken as the preferred one. It has been used successfully in "Three Gorges Project Surveying and Mapping Management System". And been proved for the ability of big data quantity processing. On a PC platform (64 M EMS memory, 200MHZ frequency), it takes only dozens of seconds to complete the computations of more than 10,000 points. The high efficiency and superiority of this algorithm in building TIN have been greatly proved.
References
- Zhang Zusun, Zhang Jianqing, Principle of Digital Photogrammetry, WuHan Technical University of Surveying and Mapping Press, 1996.
|
|
|
|