Quantized Mesh Generation : Software for Transforming Altitude Data into Tiled Representations
Lundström, Max (2023)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2023052548307
https://urn.fi/URN:NBN:fi-fe2023052548307
Tiivistelmä
This thesis presents the design and implementation of a software tool for Navielektro Ky that generates QuantizedMesh tiles from elevation data. While maps are incredibly useful for geographical analysis and visualization, achieving desirable results when utilizing them for more precise analysis and visualization can prove challenging. One solution to this problem is the use of digital elevation models (DEMs), which capture the topographical structure of the Earth’s surface. DEMs can be used to create highly accurate and detailed maps, but unfortunately, most of them are distributed in formats that themselves are not especially flexible, and require further processing for specific use cases.
QuantizedMesh tiles provide a solution to this problem by dividing the Earth’s surface into square tiles of terrain of varying levels of detail. These tiles are compressed and can be dynamically loaded and replaced during runtime, which proves especially useful in cloud-based applications, as the users only load tiles at the location and the level of detail they are interested in, resulting in lower bandwidth requirements. The use of QuantizedMesh tiles is increasingly popular in various applications, such as virtual globes, online maps, and flight simulators.
The algorithm presented in this thesis used to generate the QuantizedMesh tiles involves first creating a high-detail level of tiles and then sampling data from that layer to create the lower-detail levels. It begins with segmenting input elevation data into tile map service (TMS) grid cells. The data in these cells are then used to create QuantizedMesh tiles. In the case that the data for one TMS cell is sourced from multiple sources of elevation data, the resulting tile automatically merges both datasets to complete the tile. The resulting QuantizedMesh tiles are optimized into triangulated irregular networks (TIN) to reduce the number of redundant data points while retaining the shape of the original mesh. The tile creation algorithm scales linearly in terms of the amount of input data in relation to computing time, and allows for processing large datasets without memory issues.
The development of this tool was necessary as there are hardly any publicly available options for generating QuantizedMesh tiles that can handle the potential scale of computation required by the project stakeholders. The tiles produced by the tool can be used for various different use cases, such as line of sight calculations and terrain visualization.
QuantizedMesh tiles provide a solution to this problem by dividing the Earth’s surface into square tiles of terrain of varying levels of detail. These tiles are compressed and can be dynamically loaded and replaced during runtime, which proves especially useful in cloud-based applications, as the users only load tiles at the location and the level of detail they are interested in, resulting in lower bandwidth requirements. The use of QuantizedMesh tiles is increasingly popular in various applications, such as virtual globes, online maps, and flight simulators.
The algorithm presented in this thesis used to generate the QuantizedMesh tiles involves first creating a high-detail level of tiles and then sampling data from that layer to create the lower-detail levels. It begins with segmenting input elevation data into tile map service (TMS) grid cells. The data in these cells are then used to create QuantizedMesh tiles. In the case that the data for one TMS cell is sourced from multiple sources of elevation data, the resulting tile automatically merges both datasets to complete the tile. The resulting QuantizedMesh tiles are optimized into triangulated irregular networks (TIN) to reduce the number of redundant data points while retaining the shape of the original mesh. The tile creation algorithm scales linearly in terms of the amount of input data in relation to computing time, and allows for processing large datasets without memory issues.
The development of this tool was necessary as there are hardly any publicly available options for generating QuantizedMesh tiles that can handle the potential scale of computation required by the project stakeholders. The tiles produced by the tool can be used for various different use cases, such as line of sight calculations and terrain visualization.