A Robust Active Contour Model for Natural Scene Contour Extraction with Automatic Thresholding
Abstract
An active contour model is proposed for contour extraction of objects in complex natural scenes. Our model is formulated in an analytical framework that consists of a colour contrast metric, an illumination parameter and a blurring scale estimated using hermite polynomials. A distinct advantage of this framework is that it allows for automatic selection of thresholds under conditions of uneven illumination and image blur. The active contour is also initialised by a single point of maximum colour contrast between the object and background. The model has been applied to synthetic images and natural scenes and shown to perform well.
Cite
Text
Ngoi and Jia. "A Robust Active Contour Model for Natural Scene Contour Extraction with Automatic Thresholding." European Conference on Computer Vision, 1996. doi:10.1007/3-540-61123-1_151Markdown
[Ngoi and Jia. "A Robust Active Contour Model for Natural Scene Contour Extraction with Automatic Thresholding." European Conference on Computer Vision, 1996.](https://mlanthology.org/eccv/1996/ngoi1996eccv-robust/) doi:10.1007/3-540-61123-1_151BibTeX
@inproceedings{ngoi1996eccv-robust,
title = {{A Robust Active Contour Model for Natural Scene Contour Extraction with Automatic Thresholding}},
author = {Ngoi, Kian Peng and Jia, Jiancheng},
booktitle = {European Conference on Computer Vision},
year = {1996},
pages = {335-346},
doi = {10.1007/3-540-61123-1_151},
url = {https://mlanthology.org/eccv/1996/ngoi1996eccv-robust/}
}