2010 年 76 巻 2 号 p. 226-231
In the triple dexel model, an object shape is represented with 3 groups of axis aligned segments named X-dexels, Y-dexels, and Z-dexels. In this paper, the authors propose a data conversion method from a conventional polygonal model such as STL to a triple dexel model, and vice versa. In this method, the following technology is employed; (1) Data conversion from a polygonal model to a triple dexel model is realized by multiple applications of “depth peeling”. This process can be accelerated by using the depth buffer and stencil buffer mechanisms of GPU. (2) Data conversion from a triple dexel model to a polygonal model is realized by gathering and classifying polygons attached to the end point of neighboring dexels. An experimental conversion program is implemented and some computational experiments are performed. Our program can convert a polygonal model with 3,000,000 polygons to a triple dexel model with 1024 x 1024 resolution grids in 8 seconds. Reverse conversion from a triple dexel model to a polygonal model is realized in a second.