Browse Prior Art Database

DYNAMIC SCENE ANALYSIS

IP.com Disclosure Number: IPCOM000128468D
Original Publication Date: 1984-Jan-01
Included in the Prior Art Database: 2005-Sep-16

Publishing Venue

Software Patent Institute

Related People

Jain, Ramesh: AUTHOR [+3]

Abstract

This paper presents an overview of dynamic scene analysis. It discusses techniques for change detection, segmentation, computing optical flow and extracting information therefrom, recovering 3-Dimensional information about the objects and the motion, representation of motion in terms of low level concepts and the verbalization of motion. The emphasis is on an overview of the state of the art, rather than an exhaustive survey.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 3% of the total text.

Page 1 of 36

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

DYNAMIC SCENE ANALYSIS

Ramesh Jain

CRL-TR-6-84

THE UNIVERSITY OF MICHIGAN COMPUTING RESEARCH LABORATORY1

JANUARY 1984

Room 1079, East Engineering Building
Ann Arbor, Michigan 48109
USA
Tel: (313) 763-8000

To appear in Progress in Pattern Recognition, Vol.2, Ed. L. Kanal and A. Rosenfeld

DYNAMIC SCENE ANALYSIS

Ramesh Jain
Department of Electrical and Computer Engineering The University of Michigan
Ann Arbor MI 481092

Abstract

This paper presents an overview of dynamic scene analysis. It discusses techniques for change detection, segmentation, computing optical flow and extracting information therefrom, recovering 3-Dimensional information about the objects and the motion, representation of motion in terms of low level concepts and the verbalization of motion. The emphasis is on an overview of the state of the art, rather than an exhaustive survey.

1. Introduction

The World is dynamic. Most biological vision systems have evolved to cope with the changing world. The past decade has seen the emergence of computer vision systems.3 For a computer vision system engaged in the performance of non-trivial real world operations and tasks, the ability to cope with moving and changing objects and viewpoints is vital. Though early computer vision systems were concerned with static scenes, the last few years have seen an ever- increasing interest in computer vision systems capable of analyzing dynamic scenes.

The input to a dynamic scene analysis system is a sequence of frames of a changing world. The camera may also be moving. Each frame represents image of the scene at a particular time instant. The changes in a scene may be due to the motion of the camera or the motion of the objects, illumination changes, or the changes in the structure, size, or shape of an object. Usually it is assumed that the changes in a scene are due to motion of camera and/or objects

1 This work was supported by the National Science Foundation, under the Grant No. NSF-G-MCS-8219739. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the funding agency.

2 This work was supported by the National Science Foundation, under the Grant No NSF-G-MCS-8219739.

3 Ballard, D.H. and C. M. Brown, Computer Vision, Prentice Hall, 1982.

University of Michigan Computing Research Laboratory Page 1 Jan 01, 1984

Page 2 of 36

DYNAMIC SCENE ANALYSIS

and that the objects are either rigid or quasi-rigid; other changes are not allowed. The task of the system is to detect changes, to find motion characteristics of the observer and the objects, to recover the structure of the objects, to characterize the motion using high level abstraction, and to recognize moving objects. It is also possible that future systems will be required to observe a scene, then describe the events taking place in a language understandable by a possibl...