| Summary: | 碩士 === 國立中山大學 === 機械工程研究所 === 85 === Human vision is known to have the ability of hierarchical image processing which normally contributes to the effective processing and analysis of the received image information. Therefore, the hierarchical data structures and accompanying processing methodologies on a computer vision system now appear to be increasingly important. But on a hexagonal grid, hierarchical image data may display a rotating phenomenon, so the GBT addressing technique is introduced in this thesis to map the two-dimensional data into one dimension. Here we developed a transofrmation scheme for converting between the GBT addresses and the Cartesian coordinates using the symmetrical coordinates concepts. Via this improvement, we have eliminate the errors produced by the existing methods. An example of raster to vector conversion, using our technique, is also included in this thesis.
|
|---|
