Object oriented motion-segmentation for video-compression in the CNN-UM

T. Szirányi, Károly László, László Czúni, Francesco Ziliani

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

Object-oriented motion segmentation is a basic step of the effective coding of image-series. Following the MPEG-4 standard we should define such objects. In this paper, a fully parallel and locally connected computation model is described for segmenting frames of image sequences based on spatial and motion information. The first type of the algorithm is called early segmentation. It is based on spatial information only and aims at providing an over segmentation of the frame in real-time. Even if the obtained results do not minimize the number of regions, it is a good starting point for higher level post processing, when the decision on how to regroup regions in object can rely on both spatial and temporal information. In the second type of the algorithm stochastic optimization methods are used to form homogenous dense optical vector fields which act directly on motion vectors instead of 2D or 3D motion parameters. This makes the algorithm simple and less time consuming than many other relaxation methods. Then we apply morphological operators to handle disocclusion effects and to map the motion field to the spatial content. Computer simulations of the CNN architecture demonstrate the usefulness of our methods. All solutions in our approach suggest a fully parallel implementation in a newly developed CNN-UM VLSI chip architecture.

Original languageEnglish
Pages (from-to)479-496
Number of pages18
JournalJournal of VLSI Signal Processing Systems for Signal, Image, and Video Technology
Volume23
Issue number2
Publication statusPublished - 1999

Fingerprint

Motion Segmentation
Video Compression
Image compression
Object-oriented
Motion
Segmentation
Motion Picture Experts Group standards
MPEG-4
Locally Connected
Motion Vector
Relaxation Method
Stochastic Optimization
Stochastic Methods
Spatial Information
Parallel Implementation
Image Sequence
Post-processing
Optimization Methods
Vector Field
Chip

ASJC Scopus subject areas

  • Information Systems
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Object oriented motion-segmentation for video-compression in the CNN-UM. / Szirányi, T.; László, Károly; Czúni, László; Ziliani, Francesco.

In: Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology, Vol. 23, No. 2, 1999, p. 479-496.

Research output: Contribution to journalArticle

@article{5b7bbf23050247fabf90196c780b68d8,
title = "Object oriented motion-segmentation for video-compression in the CNN-UM",
abstract = "Object-oriented motion segmentation is a basic step of the effective coding of image-series. Following the MPEG-4 standard we should define such objects. In this paper, a fully parallel and locally connected computation model is described for segmenting frames of image sequences based on spatial and motion information. The first type of the algorithm is called early segmentation. It is based on spatial information only and aims at providing an over segmentation of the frame in real-time. Even if the obtained results do not minimize the number of regions, it is a good starting point for higher level post processing, when the decision on how to regroup regions in object can rely on both spatial and temporal information. In the second type of the algorithm stochastic optimization methods are used to form homogenous dense optical vector fields which act directly on motion vectors instead of 2D or 3D motion parameters. This makes the algorithm simple and less time consuming than many other relaxation methods. Then we apply morphological operators to handle disocclusion effects and to map the motion field to the spatial content. Computer simulations of the CNN architecture demonstrate the usefulness of our methods. All solutions in our approach suggest a fully parallel implementation in a newly developed CNN-UM VLSI chip architecture.",
author = "T. Szir{\'a}nyi and K{\'a}roly L{\'a}szl{\'o} and L{\'a}szl{\'o} Cz{\'u}ni and Francesco Ziliani",
year = "1999",
language = "English",
volume = "23",
pages = "479--496",
journal = "Journal of Signal Processing Systems",
issn = "1939-8018",
publisher = "Springer New York",
number = "2",

}

TY - JOUR

T1 - Object oriented motion-segmentation for video-compression in the CNN-UM

AU - Szirányi, T.

AU - László, Károly

AU - Czúni, László

AU - Ziliani, Francesco

PY - 1999

Y1 - 1999

N2 - Object-oriented motion segmentation is a basic step of the effective coding of image-series. Following the MPEG-4 standard we should define such objects. In this paper, a fully parallel and locally connected computation model is described for segmenting frames of image sequences based on spatial and motion information. The first type of the algorithm is called early segmentation. It is based on spatial information only and aims at providing an over segmentation of the frame in real-time. Even if the obtained results do not minimize the number of regions, it is a good starting point for higher level post processing, when the decision on how to regroup regions in object can rely on both spatial and temporal information. In the second type of the algorithm stochastic optimization methods are used to form homogenous dense optical vector fields which act directly on motion vectors instead of 2D or 3D motion parameters. This makes the algorithm simple and less time consuming than many other relaxation methods. Then we apply morphological operators to handle disocclusion effects and to map the motion field to the spatial content. Computer simulations of the CNN architecture demonstrate the usefulness of our methods. All solutions in our approach suggest a fully parallel implementation in a newly developed CNN-UM VLSI chip architecture.

AB - Object-oriented motion segmentation is a basic step of the effective coding of image-series. Following the MPEG-4 standard we should define such objects. In this paper, a fully parallel and locally connected computation model is described for segmenting frames of image sequences based on spatial and motion information. The first type of the algorithm is called early segmentation. It is based on spatial information only and aims at providing an over segmentation of the frame in real-time. Even if the obtained results do not minimize the number of regions, it is a good starting point for higher level post processing, when the decision on how to regroup regions in object can rely on both spatial and temporal information. In the second type of the algorithm stochastic optimization methods are used to form homogenous dense optical vector fields which act directly on motion vectors instead of 2D or 3D motion parameters. This makes the algorithm simple and less time consuming than many other relaxation methods. Then we apply morphological operators to handle disocclusion effects and to map the motion field to the spatial content. Computer simulations of the CNN architecture demonstrate the usefulness of our methods. All solutions in our approach suggest a fully parallel implementation in a newly developed CNN-UM VLSI chip architecture.

UR - http://www.scopus.com/inward/record.url?scp=0343193215&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0343193215&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0343193215

VL - 23

SP - 479

EP - 496

JO - Journal of Signal Processing Systems

JF - Journal of Signal Processing Systems

SN - 1939-8018

IS - 2

ER -