Most of the traditional clustering methods are based on spatial database, which reach the clustering results by operating
and computing the spatial coordinates and its attribute values in database. So it can not consider such factors of the
clustering objects as the morphological characters, topological relation etc in the process and results of clustering. From
the point of view of spatial geometry, this paper explored to build the basis of clustering analysis on graphics, divided
spatial clustering into eight types, put forward some kinds of methods of model simplification and dimension reduction
based on spatial projection and mapping, and at last demonstrated the graphics-based spatial clustering method by
introduced a fast spatial clustering method based on Graph Theory and Delaunay triangulation, which achieved good
results judged by the experiment data.
Traditional modeling methods on spatial objects are not eligible to deal well with the fuzzy features that acquired from
image, some research need to be carried out on the fuzzy spatial objects modeling. With the deep investigation on the
spatial objects model of GIS and the representation of natural geographical feature, fuzzy spatial objects have been
proposed by researchers. Referring to the characteristics of the representation of fuzzy spatial objects, a generation
method of fuzzy spatial objects based on fuzzy Neural Networks is going to be demonstrated by the authors in this paper.
By combining the fuzzy technique and neural networks, utilizing the learning ability to enhance the fuzzy membership
function and fuzzy rules, the system will be self-Adaptive. By comparing with the traditional fuzzy objects generation,
the method in this paper improves the accuracy of results according to the experiments in this paper.
With the development of computer science, many kinds of three dimensional software come to the world one after
another, and have been applied in many fields, such as PC games, military simulation and geographical information
system (GIS) and so on, among which the application of the technique of three dimensional geo-simulation in GIS field
has provided new platform, new ideas, and a new significant direction for the development of GIS. This paper is to
discuss the methods and realization of three-dimensional(3D) geosimulation in the field of bank collapse based on GIS,
with the research of bank collapse as the cut-in point, which is the important direction in the research of geological
disasters, and to give 3D simulation result of the reservoir bank, taking the related data in three gorges reservoir area as
the example, and at the same time to give the result of 3D simulation about bank collapse forecast, and at last to appraise