Video Object Segmentation of Dynamic Scenes with Large Displacements

Yinhui ZHANG; Zifen HE

doi:10.1587/transinf.2015EDL8062

Video Object Segmentation of Dynamic Scenes with Large Displacements

Yinhui ZHANG, Zifen HE

Full Text Views

0

Share
Cite this

Summary :

Segmenting foreground objects in unconstrained dynamic scenes still remains a difficult problem. We present a novel unsupervised segmentation approach that allows robust object segmentation of dynamic scenes with large displacements. To make this possible, we project motion based foreground region hypotheses generated via standard optical flow onto visual saliency regions. The motion hypotheses correspond to inside seeds mapping of the motion boundary. For visual saliency, we generalize the image signature method from images to videos to delineate saliency mapping of object proposals. The mapping of image signatures estimated in Discrete Cosine Transform (DCT) domain favor stand-out regions in the human visual system. We leverage a Markov random field built on superpixels to impose both spatial and temporal consistence constraints on the motion-saliency combined segments. Projecting salient regions via an image signature with inside mapping seeds facilitates segmenting ambiguous objects from unconstrained dynamic scenes in presence of large displacements. We demonstrate the performance on fourteen challenging unconstrained dynamic scenes, compare our method with two state-of-the-art unsupervised video segmentation algorithms, and provide quantitative and qualitative performance comparisons.

Publication: IEICE TRANSACTIONS on Information Vol.E98-D No.9 pp.1719-1723

Publication Date: 2015/09/01

Publicized: 2015/06/17

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2015EDL8062

Type of Manuscript: LETTER

Category: Image Processing and Video Processing

Authors

Yinhui ZHANG
Kunming University of Science and Technology
Zifen HE
Kunming University of Science and Technology

Keyword

video segmentation, unsupervised, saliency, motion segmentation

Cite this

Copy

Yinhui ZHANG, Zifen HE, "Video Object Segmentation of Dynamic Scenes with Large Displacements" in IEICE TRANSACTIONS on Information, vol. E98-D, no. 9, pp. 1719-1723, September 2015, doi: 10.1587/transinf.2015EDL8062.
Abstract: Segmenting foreground objects in unconstrained dynamic scenes still remains a difficult problem. We present a novel unsupervised segmentation approach that allows robust object segmentation of dynamic scenes with large displacements. To make this possible, we project motion based foreground region hypotheses generated via standard optical flow onto visual saliency regions. The motion hypotheses correspond to inside seeds mapping of the motion boundary. For visual saliency, we generalize the image signature method from images to videos to delineate saliency mapping of object proposals. The mapping of image signatures estimated in Discrete Cosine Transform (DCT) domain favor stand-out regions in the human visual system. We leverage a Markov random field built on superpixels to impose both spatial and temporal consistence constraints on the motion-saliency combined segments. Projecting salient regions via an image signature with inside mapping seeds facilitates segmenting ambiguous objects from unconstrained dynamic scenes in presence of large displacements. We demonstrate the performance on fourteen challenging unconstrained dynamic scenes, compare our method with two state-of-the-art unsupervised video segmentation algorithms, and provide quantitative and qualitative performance comparisons.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2015EDL8062/_p

Copy

@ARTICLE{e98-d_9_1719,
author={Yinhui ZHANG, Zifen HE, },
journal={IEICE TRANSACTIONS on Information},
title={Video Object Segmentation of Dynamic Scenes with Large Displacements},
year={2015},
volume={E98-D},
number={9},
pages={1719-1723},
abstract={Segmenting foreground objects in unconstrained dynamic scenes still remains a difficult problem. We present a novel unsupervised segmentation approach that allows robust object segmentation of dynamic scenes with large displacements. To make this possible, we project motion based foreground region hypotheses generated via standard optical flow onto visual saliency regions. The motion hypotheses correspond to inside seeds mapping of the motion boundary. For visual saliency, we generalize the image signature method from images to videos to delineate saliency mapping of object proposals. The mapping of image signatures estimated in Discrete Cosine Transform (DCT) domain favor stand-out regions in the human visual system. We leverage a Markov random field built on superpixels to impose both spatial and temporal consistence constraints on the motion-saliency combined segments. Projecting salient regions via an image signature with inside mapping seeds facilitates segmenting ambiguous objects from unconstrained dynamic scenes in presence of large displacements. We demonstrate the performance on fourteen challenging unconstrained dynamic scenes, compare our method with two state-of-the-art unsupervised video segmentation algorithms, and provide quantitative and qualitative performance comparisons.},
keywords={},
doi={10.1587/transinf.2015EDL8062},
ISSN={1745-1361},
month={September},}

Copy

TY - JOUR
TI - Video Object Segmentation of Dynamic Scenes with Large Displacements
T2 - IEICE TRANSACTIONS on Information
SP - 1719
EP - 1723
AU - Yinhui ZHANG
AU - Zifen HE
PY - 2015
DO - 10.1587/transinf.2015EDL8062
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E98-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2015
AB - Segmenting foreground objects in unconstrained dynamic scenes still remains a difficult problem. We present a novel unsupervised segmentation approach that allows robust object segmentation of dynamic scenes with large displacements. To make this possible, we project motion based foreground region hypotheses generated via standard optical flow onto visual saliency regions. The motion hypotheses correspond to inside seeds mapping of the motion boundary. For visual saliency, we generalize the image signature method from images to videos to delineate saliency mapping of object proposals. The mapping of image signatures estimated in Discrete Cosine Transform (DCT) domain favor stand-out regions in the human visual system. We leverage a Markov random field built on superpixels to impose both spatial and temporal consistence constraints on the motion-saliency combined segments. Projecting salient regions via an image signature with inside mapping seeds facilitates segmenting ambiguous objects from unconstrained dynamic scenes in presence of large displacements. We demonstrate the performance on fourteen challenging unconstrained dynamic scenes, compare our method with two state-of-the-art unsupervised video segmentation algorithms, and provide quantitative and qualitative performance comparisons.
ER -