Adaptive Background Estimation: Computing a Pixel-Wise Learning Rate from Local Confidence and Global Correlation Values

Mickael PIC; Luc BERTHOUZE; Takio KURITA

Adaptive Background Estimation: Computing a Pixel-Wise Learning Rate from Local Confidence and Global Correlation Values

Mickael PIC, Luc BERTHOUZE, Takio KURITA

Full Text Views

0

Share
Cite this

Summary :

Adaptive background techniques are useful for a wide spectrum of applications, ranging from security surveillance, traffic monitoring to medical and space imaging. With a properly estimated background, moving or new objects can be easily detected and tracked. Existing techniques are not suitable for real-world implementation, either because they are slow or because they do not perform well in the presence of frequent outliers or camera motion. We address the issue by computing a learning rate for each pixel, a function of a local confidence value that estimates whether a pixel is (or not) an outlier, and a global correlation value that detects camera motion. After discussing the role of each parameter, we report experimental results, showing that our technique is fast but efficient, even in a real-world situation. Furthermore, we show that the same method applies equally well to a 3-camera stereoscopic system for depth perception.

Publication: IEICE TRANSACTIONS on Information Vol.E87-D No.1 pp.50-57

Publication Date: 2004/01/01

Publicized

Online ISSN

DOI

Type of Manuscript: Special Section PAPER (Special Section on Machine Vision Applications)

Category: Background Estimation

Cite this

Copy

Mickael PIC, Luc BERTHOUZE, Takio KURITA, "Adaptive Background Estimation: Computing a Pixel-Wise Learning Rate from Local Confidence and Global Correlation Values" in IEICE TRANSACTIONS on Information, vol. E87-D, no. 1, pp. 50-57, January 2004, doi: .
Abstract: Adaptive background techniques are useful for a wide spectrum of applications, ranging from security surveillance, traffic monitoring to medical and space imaging. With a properly estimated background, moving or new objects can be easily detected and tracked. Existing techniques are not suitable for real-world implementation, either because they are slow or because they do not perform well in the presence of frequent outliers or camera motion. We address the issue by computing a learning rate for each pixel, a function of a local confidence value that estimates whether a pixel is (or not) an outlier, and a global correlation value that detects camera motion. After discussing the role of each parameter, we report experimental results, showing that our technique is fast but efficient, even in a real-world situation. Furthermore, we show that the same method applies equally well to a 3-camera stereoscopic system for depth perception.
URL: https://globals.ieice.org/en_transactions/information/10.1587/e87-d_1_50/_p

Copy

@ARTICLE{e87-d_1_50,
author={Mickael PIC, Luc BERTHOUZE, Takio KURITA, },
journal={IEICE TRANSACTIONS on Information},
title={Adaptive Background Estimation: Computing a Pixel-Wise Learning Rate from Local Confidence and Global Correlation Values},
year={2004},
volume={E87-D},
number={1},
pages={50-57},
abstract={Adaptive background techniques are useful for a wide spectrum of applications, ranging from security surveillance, traffic monitoring to medical and space imaging. With a properly estimated background, moving or new objects can be easily detected and tracked. Existing techniques are not suitable for real-world implementation, either because they are slow or because they do not perform well in the presence of frequent outliers or camera motion. We address the issue by computing a learning rate for each pixel, a function of a local confidence value that estimates whether a pixel is (or not) an outlier, and a global correlation value that detects camera motion. After discussing the role of each parameter, we report experimental results, showing that our technique is fast but efficient, even in a real-world situation. Furthermore, we show that the same method applies equally well to a 3-camera stereoscopic system for depth perception.},
keywords={},
doi={},
ISSN={},
month={January},}

Copy

TY - JOUR
TI - Adaptive Background Estimation: Computing a Pixel-Wise Learning Rate from Local Confidence and Global Correlation Values
T2 - IEICE TRANSACTIONS on Information
SP - 50
EP - 57
AU - Mickael PIC
AU - Luc BERTHOUZE
AU - Takio KURITA
PY - 2004
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E87-D
IS - 1
JA - IEICE TRANSACTIONS on Information
Y1 - January 2004
AB - Adaptive background techniques are useful for a wide spectrum of applications, ranging from security surveillance, traffic monitoring to medical and space imaging. With a properly estimated background, moving or new objects can be easily detected and tracked. Existing techniques are not suitable for real-world implementation, either because they are slow or because they do not perform well in the presence of frequent outliers or camera motion. We address the issue by computing a learning rate for each pixel, a function of a local confidence value that estimates whether a pixel is (or not) an outlier, and a global correlation value that detects camera motion. After discussing the role of each parameter, we report experimental results, showing that our technique is fast but efficient, even in a real-world situation. Furthermore, we show that the same method applies equally well to a 3-camera stereoscopic system for depth perception.
ER -