| In recent years, in order to solve theproblems of resources, environment, ecology, population, economic and social and to construct harmonious society, our country opened the First Geographical Conditions Census in 2013. In this paper, based on the main job content of geographical conditions census project, we extracted road information from geographical conditions census image for high resolution remote sensing image of Yanzhou. On the experimental platform of eCongnition8.9, we were carried out the following experiment and research:a. First of all, we study the main content and technical route of geographic conditions census, and to study the classification and collection requirements of the road in the census standard. This paper summarizes the classification of roads in the census, and the basic characteristics in image of different types of roads were analyzed and summarizedthe different characteristics of the census road extraction.b. This paper studies the major technical theories and methods of object-oriented to road extraction; and for the characteristics of road in geographical conditions census, proposed screening process for rural roads in large contiguous areas.c. For different categories of road in geographic conditions census, we conducted experiments using multi-scale segmentation of eCognition8.9. For the comparative analysis of the experimental results, we obtained the optimal segmentation scale of four road types: the optimal segmentation scale of railways, highways, urban roads and rural roads are 105,75,70 and 100.d. Extracted information of roads for different types of roads in experimental zones, based on object-oriented, and screened rural roads extraction results by the feature; then compared quality evaluation of the roads based elements and analyzed the precision of roads. Conclusions indicate that, the road information extraction results based on eCognition met the technical requirements of the Geographical Conditions Census basically. |
PDF Full Text Request