Research On Inhabit Extraction Based On Hierarchical Classified Means

There is strong space correlativity between the space distributing of residential place and that of regional population, economy, property . It provides basic framework for the spacial extensity of statistic data for social economy such as population, economy, property, etc.. The development of remote sensing technology provides advanced means for the acquirement of the information of residential area.With the principle of the least extraction inaccuracy of residential area and with the aim of improving extraction accuracy of residential area, the paper which is based on the remote sensing data of SPOT5-HI studies at the extraction residential area through hierarchical classified thinking and studies at the mixed and classified phenomena between residential area and non-residential area. The paper fully analyses characteristic distributing of the earth surface, organizes messed distributing of the earth surface, and builds up a type of hierarchical classified tree according to the result of field survey which conforms to the hierarchical classified thinking, what's more, in order to gain the information of residential area, the paper builds up an appropriate model of layered extraction in accordance with the aim of research, and goes on with mixed analysis between characteristic of the earth surface and residential area according to the information of residential area. Finally, the information of residential area are achieved. The main study conclusions are as follows.l,We disassemble complex information of the earth surface layer upon layer at research area, go on with logical estimation orderly through hierarchical classified thinking, dig out and differentiate the characteristic wave band of information of residential area and non-residential area, combine or operate wave band effectively , and build up classified instrument according to every layer. We learn from the result of every layer's classified precision estimation that the pertinence of the method is very strong; In order to avoid the redundancy of data, we unite and adjust reversely the earth objects which are distributing at research area; In order to achieve pure information of residential area, we sample and analyze the information of non-residential area ,dig out and differentiate the judge basis of the information of residential area and non-residential area, and separate gradually the information of non-residential area from residential area after several times' mixed study at the information between residential area and non-residential area. The classified precision of residential area which is achieved at last is 79%. Compared to the known extraction result of residential area, we find that the detail information of the earth enlarges with the increasing of resolution and it is more difficult to extract the information of residential area. However, precision is not improving evidently. The other reason of affecting classified precision is the spacial resolution of estimating precision data. It is natural that the extraction precision of residential area will decrease when estimate precision based on pixels through spacial resolution which is higher than classified data. We found that the study at extraction of complex residential area is effective with the hierarchical classified thinking by analyzing synthetically all the steps of extraction studying and the classified precision which is achieved in the end.2,The paper studies at the extraction of roads information which strongly affects the extraction precision of the residential area according to the Marr 's ocular principle. The researchon three steps which include preliminary transaction of roads, the lower layered extraction of roads and the late transaction of roads is directly related with the extraction accuracy of roads information. The paper based on the result of former research extracts the roads information by adopting the preliminary transaction of automatic filtering the waves, ISODATA fuzzy clustering and the late transaction of roads based on geometrical estimation.