Background modeling and subtraction by codebook construction

We present a new fast algorithm for background modeling and subtraction. Sample background values at each pixel are quantized into codebooks which represent a compressed form of background model for a long image sequence. This allows us to capture structural background variation due to periodic-like...

Full description

Saved in:
Bibliographic Details
Published in2004 International Conference on Image Processing, 2004. ICIP '04 Vol. 5; pp. 3061 - 3064 Vol. 5
Main Authors Kyungnam Kim, Chalidabhongse, T.H., Harwood, D., Davis, L.
Format Conference Proceeding
LanguageEnglish
Published Piscataway NJ IEEE 2004
Subjects
Applied sciences
Artificial intelligence
Brightness
Computer science; control theory; systems
Computer vision
Educational institutions
Exact sciences and technology
Gaussian processes
Image coding
Information technology
Layout
Lighting
Pattern recognition. Digital image processing. Computational geometry
Pixel
Robustness
Luminance
Codebook
Subtraction
Image processing
Image sequence
Video signal
Illumination
Shadow
Periodic motion
Fast algorithm
Modeling
Online AccessGet full text
ISBN0780385543
9780780385542
ISSN1522-4880
DOI10.1109/ICIP.2004.1421759

Cover

More Information
Summary:We present a new fast algorithm for background modeling and subtraction. Sample background values at each pixel are quantized into codebooks which represent a compressed form of background model for a long image sequence. This allows us to capture structural background variation due to periodic-like motion over a long period of time under limited memory. Our method can handle scenes containing moving backgrounds or illumination variations (shadows and highlights), and it achieves robust detection for compressed videos. We compared our method with other multimode modeling techniques.
ISBN:0780385543
9780780385542
ISSN:1522-4880
DOI:10.1109/ICIP.2004.1421759