Finding boundaries in natural images: A new method using point descriptors and area completion
Research output: Contribution to journal › Conference article › Research › peer-review
We develop an approach to image segmentation for natural scenes containing image texture. One general methodology which shows promise for solving this problem is to characterize textured regions via their responses to a set of filters. However, this approach brings with it many open questions, including how to combine texture and intensity information into a common descriptor and how to deal with the fact that filter responses inside textured regions are generally spatially inhomogeneous. Our goal in this paper is to introduce two new ideas which address these open questions and to demonstrate the application of these ideas to the segmentation of natural images. The first idea consists of a novel means of describing points in natural images and an associated distance function for comparing these descriptors. This distance function is aided in textured regions by the use of the second idea, a new process introduced here which we have termed area completion. Experimental segmentation results which incorporate our proposed approach into the Normalized Cut framework of Shi and Malik are provided for a variety of natural images.
Original language | English |
---|---|
Journal | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
Pages (from-to) | 751-766 |
Number of pages | 16 |
ISSN | 0302-9743 |
DOIs | |
Publication status | Published - 1998 |
Externally published | Yes |
Event | 5th European Conference on Computer Vision, ECCV 1998 - Freiburg, Germany Duration: 2 Jun 1998 → 6 Jun 1998 |
Conference
Conference | 5th European Conference on Computer Vision, ECCV 1998 |
---|---|
Country | Germany |
City | Freiburg |
Period | 02/06/1998 → 06/06/1998 |
Bibliographical note
Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1998.
ID: 302060581