Project Title: Performance
evaluation of shape simplification methods
Student: Kennant Kin Lun
Tom
Supervisor: Mrs. Gisela
Klette
Co-supervisor: Prof. Reinhard
Klette
Brief Description of the
project:
The aim of my project is to investigate different algorithms
for shape simplifications of 2D components in digital images.
Thinning algorithms are characterized by an iterative process of transforming
border pixels into background pixels of components of digital images. Some
algorithms stop before the resulting subset is a set of digital arcs in order
to preserve connectivity. Ideal skeletons are sets of digital arcs or curves.
These skeletons are very useful for classifications in image analysis. I
implemented three different thinning algorithms in order to evaluate the
performance (e.g. execution time) and the quality of resulting images in terms
of connectivity, noise sensitivity and representation of shape. I also
implemented several supporting tools (e.g. to identify border pixels in a
digital image, to increase the grid resolution, applets for the visualisation
of the iterative thinning etc.)
Distance skeletons are based on distance transformations. They represent
unconnected subsets of a given component of a digital image. The original
component is reconstructable from the distance skeleton. This property makes
distance skeletons useful for data compression.
All algorithms and supporting tools are implemented in JAVA.
An example is shown below. The image on the left is the original image and the
image on the right is the result of a sequential thinning algorithm as
described in my report in details.
Input Image Output Image