Tracking 3D human body using particle filter in moving monocular camera

Sungmin Kim, Chang Beom Park, Seong Whan Lee

Research output: Contribution to journalConference article

11 Citations (Scopus)

Abstract

In this paper, we propose a method for human tracking using 3D human body model in a video sequence with a monocular moving camera. Tracking a human with unconstrained movement in moving monocular camera image sequence is extremely challenging. Our 3D human body model which is formed with articulation model of hierarchical tree structure can express all human's movement by parameters. We can obtain 3D human body model which has the most similar shape with input image through similarity matching. In order to predict the region and movement of human using 3D human body model in the obtained current frame, we use the particle filter which predicts the posterior distribution by the random probability variable based on Monte Carlo sampling. As a result, it can be possible to track robustly for human 's motion and random movement of camera in the environment with moving camera. We can get the result of converging toward minimized error values using boundary distance between a predicted 3D human body model and an input image. In the result of experiment, the proposed method showed correct tracking result for complex background and various human movements.

Original languageEnglish
Article number1699962
Pages (from-to)805-808
Number of pages4
JournalProceedings - International Conference on Pattern Recognition
Volume4
DOIs
Publication statusPublished - 2006
Event18th International Conference on Pattern Recognition, ICPR 2006 - Hong Kong, China
Duration: 2006 Aug 202006 Aug 24

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Fingerprint Dive into the research topics of 'Tracking 3D human body using particle filter in moving monocular camera'. Together they form a unique fingerprint.

  • Cite this