| |
|
|
| |
| Abstract |
1. Student data and contact requisites.
Surname and name: Abaev Sergey Rashitovich.
Place of study: Donetsk national technical university (DonNTU), the master of computer facilities and computer science faculty.
Group: CEM-01ì.
Direction of education: computer sciences.
Scientific headmaster: Averin Gennadiy Viktorovich.
Cathedra: CSM.
2. Theme of masters work.
The Substantiation of methods for objects identification on space photos of city territories.
3. The summary of work.
In the modern world topographical cards give us representation about the world surrounding us and allow us to be guided in it, showing all visible elements of district with an identical details. On them are displayed: the relief, hydrography, vegetation, ground and sand, settlements, a roadway system, social and economic and other objects that allows to estimate territory complexly.
At present on the world map practically does not remain white spots - topographical cards cover almost all surface of a land of our planet, though not all of them equally detailed. However the biggest part of these maps is not updated regularly. Meanwhile, the modern world very changeble: cities grow, new settlements appears, roads builds, communication networks, engineering constructions, new areas of extraction of minerals accustom, woods are cut down, the structure of land tenure changes. Therefore constantly appears a problem of updating our topographical maps.
As a primary material for topographical maps traditionally used aerial photographs. Space digital pictures open new opportunities: reduction in price of retakes, increase in the area of scope of district and decrease in the distortions connected with a relief. Besides generalization of the image on small-scale maps becomes simpler: instead of labour-consuming simplification of large-scale maps it is possible to use space pictures of the average resolution. Therefore shootings from space are used more and more widely, and in the long term can become the basic method of updating topographical maps.
But for high-grade work with maps it is not enough to have simply exact pictures. For performance of manipulations with maps in modern geoinformation systems, it is necessary also to be able to allocate cartographical objects in pictures (structures, roads, forest belts, the rivers, etc.), for their subsequent transformation to digital objects.
Proceeding from the aforesaid, the theme of work directed on the analysis of city territories space pictures, with the subsequent allocation of objects is actual.
4. Description of masters work.
Problem on which decision master work is directed.
Work is directed on the decision of a choice problem of optimum algorithms or their combinations for qualitative allocation of objects in space pictures of city territories.
The purpose of work.
Reception of comparative characteristics of recognition algorithms of objects, their analysis and substantiation of their usefulness.
Subject and object of researches.
An object of research are an algorithms of objects allocation oin space pictures of city territories.
Object of research - space pictures of Donetsk territory.
Problems of work.
1. Search of existing allocation of objects algorithms on space pictures of city territories.
2. Classification, the analysis and an estimation of existing algorithms quality of objects allocation.
3. Creation of test software product, for a visual estimation of quality of existing algorithms of allocation of objects.
4. A substantiation of algorithms application methods of objects identification and their hybrids.
The idea of work consists in a mathematical substantiation of practical algorithms of objects allocation use on space pictures of city territories.
Expected results of work.
The given development should define the most suitable algorithms yielding good results of objects recognition on space pictures of city territories.
Existing researches and developments.
This theme at the moment publicly practically is not opened. All developments are conducted in the closed institutes because an aspects of the state importance are mentioned. It is necessary to note even that fact, that in no one of modern geo-information modelling packages there is no module allowing to distinguish pictures of city territories though it considerably would increase their efficiency.
|
|
|
|
