A Model of Sounding Generalization Based on Recognition of Terrain Features

As a main part of nautical chart generalization, sounding generalization is also one of the bottlenecks in the way of automatic chart generalization. Because submarine topography is mostly represented by sounding on chart, to some extent, generalization of chart mainly means sounding selection. Even in manual generalization, it is one of the most difficult and time-consuming links in the production chain, not to mention in the automated charting procedures. In this paper, author proposed a new theoretical model, which based on recognition of topographic features. With the aid of Delaunay triangulation, the information, which contains the topology, distribution character, distribution density, can all be gained. All the methods mentioned in the paper have been proved to be feasible and sound. 1. Rules of Sounding Selection The major rules for sounding selection include:  Soundings which are at the top of a raised under-water area (shoals, rocks, etc.) have the highest priority to be selected.  Select the soundings, which show the waterways.  Select the soundings, which represent the outline of sea, bottom the best.  The density of selected soundings must increase in shoal areas and decrease in less dangerous areas.  In areas of homogeneous depth, soundings fulfilling the chart should formas far as possiblea rhomboidal pattern. Obviously, we can conclude from above, sounding selection deals with synchronously both spatial and attributive characteristics of chart features. There are different rules for different areas as well as different kinds of terrain features. That only have we understood them fully, the proper and efficient automatic generalization could be realized. So this paper proposes an important idea, that recognition of terrain is the basis of automatic sounding selection. 2. Recognition of Topographic Features In order to extract the terrain information from those discrete soundings, this paper proposes a model based on sounding clustering and sounding tree, with the aid of Delaunay triangulation. 2.1 Sounding Clustering The aim of space clustering is to divide the whole object into several groups, according to different standards, and use it as the basis of corresponding analysis. It is just the united clustering, including attribute data (depth) and geometry data, that sounding clustering is. Definition 1: Sounding set can be expressed by set function ) , , ( 2 1 n p p p P   , if , and P z y x p i i i i   ) , , (   j i p p  , j i  , then may be called as the sounding cluster, its coordinate and depth are expressed by , and . P