clustering, dimensionality reduction, image segmentation

In our research, we concentrate on different aspects of the dimensionality reduction problem. One of its practical implementations is the image segmantation problem, that is a task to find several parts of a grayscale image that belong together (form one object). We present a segmentation algorithm that utilizes three fuzzy transform components for selecting suitable subareas of the image and then the well-known k means clustering is applied.