Segmentation of moving objects by robust motion parameter estimation over multiple frames

Ayer, Steven K.; Schroeter, P.; Bigün, Josef

doi:10.1007/bfb0028364

Cited by 63 publications

(38 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are two main approaches to determining the contents of the layers, of which the dominant motion approach (e.g., [1], [12], [13], [14], [15]) is the most straightforward. Here, a single motion is robustly fitted to all pixels, which are then tested to see whether they really fit that motion (according to some metric).…”

Section: Review Of Previous Workmentioning

confidence: 99%

“…The pixels which agree with the motion are labeled as being on that layer. At this stage, either this layer can be labeled as "background" (being the dominant motion), and the outlier pixels as belonging to foreground objects [14], [15], or the process can be repeated recursively on the remaining pixels to provide a full set of layers for further analysis [1], [12], [13].…”

Section: Review Of Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

Layered motion segmentation and depth ordering by tracking edges

Smith

Drummond

Cipolla

2004

IEEE Trans. Pattern Anal. Machine Intell.

View full text Add to dashboard Cite

Abstract-This paper presents a new Bayesian framework for motion segmentation-dividing a frame from an image sequence into layers representing different moving objects-by tracking edges between frames. Edges are found using the Canny edge detector, and the Expectation-Maximization algorithm is then used to fit motion models to these edges and also to calculate the probabilities of the edges obeying each motion model. The edges are also used to segment the image into regions of similar color. The most likely labeling for these regions is then calculated by using the edge probabilities, in association with a Markov Random Field-style prior. The identification of the relative depth ordering of the different motion layers is also determined, as an integral part of the process. An efficient implementation of this framework is presented for segmenting two motions (foreground and background) using two frames. It is then demonstrated how, by tracking the edges into further frames, the probabilities may be accumulated to provide an even more accurate and robust estimate, and segment an entire sequence. Further extensions are then presented to address the segmentation of more than two motions. Here, a hierarchical method of initializing the Expectation-Maximization algorithm is described, and it is demonstrated that the Minimum Description Length principle may be used to automatically select the best number of motion layers. The results from over 30 sequences (demonstrating both two and three motions) are presented and discussed.

show abstract

Section: Review Of Previous Workmentioning

confidence: 99%

Section: Review Of Previous Workmentioning

confidence: 99%

Layered motion segmentation and depth ordering by tracking edges

Smith

Drummond

Cipolla

2004

IEEE Trans. Pattern Anal. Machine Intell.

View full text Add to dashboard Cite

show abstract

“…Bouthemy and Francois [9] treated the problem of independent motion detection as a statistical regularization problem and attempted to use the Markov Random Field model to solve for the problem. Ayer et al [7] used robust statistical regression techniques to detect independent motion. Smith and Brady [42] used geometric constraints for independent motion segmentation.…”

Section: Motivations and Related Workmentioning

confidence: 99%

Qualitive Detection of Independently Moving Targets in MPEG Video Within the Compressed Domain

Zhang¹

2004

View full text Add to dashboard Cite

The public reporting burden for this collection of information is estimated to average 1 hour per response, including gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comment information, including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Sery 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwith penalty for failing to comply with a collection of information if it does not display a curretitly valid 0MB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. < ^5^ REPORT DATE IDD-MM-YYYY)2. REPORT TYPE This is the final report to AFOSR and AFRL/IFEC for the project Qualitative Detection of Independently Moving Targets in MPEG Video within the Compressed Domain. This project focuses on developing theory and techniques for large scale surveillance video data summarization and unsupervised segmentation of video streams regarding whether there is a presence of independent motion in the streams. We propose a holistic, in-compression approach to efficient video prostanding. By efficient, we mean that the processing speed is close to or even faster than real-time in "normal" platforms(we do not assimie using special hardware or any parallel machines) while still maintaining comparable quality with stateof-the-art methods. By prostanding, we mean to aim at those tasks that are between the traditional video processing and traditional video understanding. We target surveillance appKcations. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)Research Fund of State University of New York P. 0. 6000 Binghamton, NY 13902-6000 PERFORMING ORGANIZATION REPORT NUMBER SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)Air SPONSOR/MONITOR'S ACRONYM(S) AFOSR SPONSOR/MONITOR'S REPORT NUMBER(S) DISTRIBUTION/AVAILABILITY STATEMENTDistribution Statement A. Approved for public release; distribution is unlimited. SUPPLEMENTARY NOTES ABSTRACTThis is the final report to AFOSR and AFRL/IFEC for the project Qualitative Detection of Independently Moving Targets in MPEG Video within the Compressed Domain. This project focuses on developing theory and techniques for large scale surveillance video data summarization and xmsupervised segmentation of video streams regarding whether there is a presence of independent motion in the streams. We propose a holistic, in-compression approach to efficient video prostanding. By efficient, we mean that the processing speed is close to or even faster than real-time in "normal" platforms(we do not assume using special hardware or any parallel machines) while still maintaining comparable quality with stateof-the-art methods. By prostanding, we mean to aim at those tasks that are between the traditional video processing and traditional video understanding. We target surveillance applications. This project focuses on developing theory and techniques for large scale surveillance video data summarization and unsupervised segmentation of video str...

show abstract

“…Other related research has worked on developing target motion estimators, e.g. [22,23,24,25]. These, however, are focused more on estimating models (usually smooth, periodic or planar models) of the moving targets.…”

Section: Background and Constraintsmentioning

confidence: 99%