Unsupervised image segmentation using robust clustering
Date of Issue1999
School of Electrical and Electronic Engineering
Unsupervised image segmentation is widely recognized as a difficult and challenging computer vision problem, in which, the fundamental issue is that of determination of the number and boundaries of different regions without a priori knowledge. There are two parts of work in this thesis. In the first part, image segmentation is addressed as detecting the boundaries between different regions, where edge flow is successfully investigated and implemented in solving both texture and color image segmentation. Edge flow is used to identify and integrate the direction of change in various types of image attributes at each image location and segment the image using a heuristic post-processing.
DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing