Texture
Segmentation Example
As a basic test for the segmentation step of our algorithm, we took
synthetic
compositions of gray-scaled textures from Thomas Hofmann, Jan Puzicha,
and
Joachim M. Buhmann, "Unsupervised texture segmentation in a
deterministic
annealing framework" paper. Each of the compositions consists of five
different
textures. We applied our segmentation algorithm with windows of 32x32
pixels
size, grid shift of 8 pixels, L=3 wavelet decomposition levels with
rbio_3.3
and db4 wavelet filters. Both filters provided us with similar results
shown
below:
Figure 6: Texture Segmentation Example. Top
left
- the original image consist-ing of 5 regions, each of different
texture.
The rest five images are the obtained segmentation of the original
image
into 5 segments.
