| Popis: |
This paper first synthetically analyzes three classical algorithms of constructing Delaunay Triangulation Irregular Net (DTIN), then uses improved convex hull algorithm to construct non-constrained D-TIN. The non-constrained D-TIN may cause some Delaunay triangles which span contour lines, so these Delaunay triangles must be dealt with. Then the local adjusting algorithm based on diagonals changing of the impacted region is used to embed contour lines into D-TIN as constrained segments and ultimately construct constrained D-TIN (CD-TIN). After constructing CD-TIN which supports the spatial data mining, it is easy to compute normal vectors of Delaunay triangles in the CD-TIN, and then get their gradient values. By means of the relationship of gradient values and terrain features, flats and mountain regions can be distinguished from topographic map based on contour lines. Even the terrain features such as mountain peak can be also recognized in the recognized mountain regions primarily by Flat-Triangle whose three vertexes have the same digital elevation. |
