Boyko V. Development and research of statistical classification methods of geographic images textures

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0410U004720

Applicant for

Specialization

  • 05.13.06 - Інформаційні технології

28-10-2010

Specialized Academic Board

K 08.051.01

Essay

The object of research is processes of classification of images of a terrestrial surface in space pictures of the high permission on their textural component. An object of research are methods and information technology of classification of space pictures of the high permission on the basis of research of their textural component. Research methods: methods of orthogonal and integrated transformations, statistical methods of recognition of images, methods of processing of images, methods of remote sounding of the Earth, object-oriented programming and computer graphics. The method of classifications of the image of structures in new basis which is received by ortogonalition of histogram characteristics of reference structures with use by Gram-Shmidta transformation is offered. The method of classification of structures of space pictures in spectral basis singular vectors of multidimensional base reference of histogram characteristics which are used in the presence of several reference images of structures of each class is offered. The structure characteristic, invariant to influence of turn of the image which is based on transformations to spectral area is offered. The new information technology is developed for the decision of a wide range of problems of wildlife management on the basis of a combination of the developed methods of classification of the image of structures and the offered spectral characteristics.

Files

Similar theses