Feature Selection and Analysis for a

Content-Based Image Database

Funded by National Science Foundation: IRI-9529045

Methods are being developed to extract both low-level and high-level image features for use in indexing and retieval of images and videos. Images from two specific application areas are being considered: color photographs and skin cancer images. First the geometry and color of images are corrected, then they are segmented to regions, each corresponding to an object or a part of it. Finally, the properties of the image regions are quantified to derive the image features.

PI: Ardy Goshtasby,

Collaborators: Clement Yu, UIC; Steve Bines, M.D. and David Roseman, M.D., Rush Med. Ctr., Chicago, IL.

An Example Using an Image of a Skin Lesion:

Original Image and Color Corrected Image

Image is segmented to extract the lesion boundary.

A Lesion bounday is characterized by contour segments forming it.

[WSU Home Page] [CSE Department Home Page] [Intelligent Systems Lab Page]

For more information contact: A. Goshtasby (ardeshir@cs.wright.edu).

Last modified: 8/27/98.