Yüksek Lisans Tezi: Yırtılabilir Zar Modeli Yardımıyla Görüntü Sıkıştırma
Istanbul Technical University, Institute of Sciences and Technology, Aug. 1995
Abstract:
In this thesis, a regularized approach to image coding is presented.
In this approach, both coding part and decoding part are based on the regularized
solutions of the weak membrane modeling.
Boundary information of objects of an image, form a sparse data which
is one of the most important representation of that image. Boundaries are
the regions where sharp changes occur in pixel intensities. Regions except
these discontinuities are smooth which means pixel values don't represent
sharp changes. This property allows to determine the pixel values of smooth
regions by using data along the two sides of edges which present at discontinuities.
In a two dimensional image, the data corresponding the values of pixels
neighboring to the edge locations produce a very sparse image. By using
an appropriate surface reconstruction algorithm, a dense image similar
to the original one can be reconstructed.
In both coding and decoding parts, weak membrane model which is a surface
reconstruction algorithm, is used to eliminate noise, detect edges and
obtain the dense image from sparse edge information. Weak membrane model
preserves discontinuities while smoothing the regions apart from boundaries.
This model is expressed by a non-convex energy functional representing
the behaviour of a membrane.
The method is applied to various synthetic and real images. The compression
ratio for syntetic images is approximately 40:1 and for real images approximately
20:1.