A Pattern Matching Technique for Detecting Similar 3D Terrain Segments

3D terrain data have been used for various systems such as navigation and landscape analysis systems. Recently, the number of 3D terrain data and databases is increasing rapidly. To handle these data effective, such systems need the ability to search 3D terrains to find matches based on shape similarity. In this paper, we propose a pattern matching technique for searching similar terrain segments from databases of 3D terrain. Since 3D terrain data are composed of three dimensional data, pattern matching computation costs are extremely high. One of the most important processes for pattern matching is the extraction of shape features from the target 3D terrain data, which requires a significant number of computations. In our system, we have applied HLAC (Higher Order Local Autocorrelation) shape features for detecting similar 3D terrain segments. The use of HLAC shape features enables fast scanning and extraction of 3D shape features by reflecting the shift invariant properties of the HLAC. Our preliminary search system successfully detects similar 3D terrain segments from databases of 3D terrains. Our pattern matching technique can be applied to various fields of research including geographical data classification and landscape visualizations. Key-Words: HLAC, Similarity Retrieval, Pattern Matching, Voxel, Terrain