1. Science
  2. Publications
  3. Information Processing Systems
  4. 1(117)'2014
  5. Segmentation and compression of realistic images on the basis of contour-form of informing areas

Segmentation and compression of realistic images on the basis of contour-form of informing areas

V.G. Ivanov, Y.V. Lomonosov, M.G. Lyubarskiy
Annotations languages:

Modification of method is in-process examined allowing to make processing of realistic images with dividing of all of plane of image into informing areas is an object and indeterminant is a background. Image segmentation is executed inplane вейвлет-коэффициентов of the first level of decomposition with the use of methods of konturizacii veyvlet areas of the not equal corrupting filters got as a result of application. This segmentation is conducted with the purpose of exposure of the most informing areas images to which in future at a compression the low «weak» coefficients of quantum are used. For other areas of image more high «strong» coefficients of quantum are used. As a result of the differentiated going near treatment of the got areas of image the general increase of degree of compression without the visible losses of quality is arrived at in the most informing areas. The special attention is spared possibility of receipt of more exact and thin structure of contour, dividing an image into informing and not informing areas. The basic idea of forming of more exact contour of division of image on different areas is a search and exception from the informing areas of such вейвлет-коэффициентов which have high-frequency of reiteration in area of background. Quantitative descriptions of degree of compression of image are got during his segmentation with the specified contour of division of areas.
Keywords: image segmentation, wavelet-analysis, compression of data