Learning a classification model for segmentation

DMCA

Learning a classification model for segmentation (2003)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Xiaofeng Ren , Jitendra Malik

Venue:	In Proc. 9th Int. Conf. Computer Vision
Citations:	281 - 2 self

BibTeX

@INPROCEEDINGS{Ren03learninga,
    author = {Xiaofeng Ren and Jitendra Malik},
    title = {Learning a classification model for segmentation},
    booktitle = {In Proc. 9th Int. Conf. Computer Vision},
    year = {2003},
    pages = {10--17}
}

OpenURL

Abstract

We propose a two-class classification model for grouping. Human segmented natural images are used as positive examples. Negative examples of grouping are constructed by randomly matching human segmentations and images. In a preprocessing stage an image is oversegmented into superpixels. We define a variety of features derived from the classical Gestalt cues, including contour, texture, brightness and good continuation. Information-theoretic analysis is applied to evaluate the power of these grouping cues. We train a linear classifier to combine these features. To demonstrate the power of the classification model, a simple algorithm is used to randomly search for good segmentations. Results are shown on a wide range of images. 1.

Keyphrases

classification model two-class classification model human segmentation preprocessing stage natural image positive example linear classifier wide range good continuation classical gestalt cue simple algorithm information-theoretic analysis negative example good segmentation

DMCA