Presentation 2004-12-17
Abnormality Detection for an Image-based Surveillance Robot
Kentaro YOKOI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose an image-based surveillance robot which can detect abnormalities, such as unattended baggage Compared to surveillance systems with fixed cameras surveillance robots can monitor a large area with fewer cameras. However, the robots experience difficulties in abnormality detection because of intervals between captured images To detect abnormalities from these images a method for accurate image position adjustment and abnormality detection tolerant to these intervals is required. We propose i) a method for the image position adjustment based on perspective projection and ii) nonparametric abnormality detection based on posterior probability of sample patterns. These methods have been tested using a movable camera within an office environment. The system can adjust image positions containing 10~20cm error relative to the surveillance subjects at l~2m distance from the camera and correctly detect abnormalities in a scene with complex textures where former methods, such as background subtraction and subspace methods, fail.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) video surveillance / abnormality detection / surveillance robot / nonparametric method
Paper # PRMU2004-144
Date of Issue

Conference Information
Committee PRMU
Conference Date 2004/12/10(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Abnormality Detection for an Image-based Surveillance Robot
Sub Title (in English)
Keyword(1) video surveillance
Keyword(2) abnormality detection
Keyword(3) surveillance robot
Keyword(4) nonparametric method
1st Author's Name Kentaro YOKOI
1st Author's Affiliation Corporate Research and Development Center, TOSHIBA Corporation,()
Date 2004-12-17
Paper # PRMU2004-144
Volume (vol) vol.104
Number (no) 524
Page pp.pp.-
#Pages 6
Date of Issue