Title :
Pedestrian Segmentation From Uncalibrated Monocular Videos Using a Projection Map
Author :
Jo, Younggwan ; Nam, Woonhyun ; Han, Joon Hee
Author_Institution :
Dept. of Comput. Sci. & Eng., Pohang Univ. of Sci. & Technol., Pohang
fDate :
7/1/2009 12:00:00 AM
Abstract :
We present a new method for segmenting the foreground region of the image of multiple pedestrians from monocular surveillance videos. This method requires neither camera calibration nor planar ground assumption. The size and orientation of a pedestrian projection are estimated at each image point and registered in a pedestrian projection map. Individual pedestrians are segmented from the foreground region of input images using an expectation maximization (EM) algorithm and the constructed map.
Keywords :
expectation-maximisation algorithm; image registration; image segmentation; video signal processing; video surveillance; expectation maximization algorithm; foreground region segmentation; image registration; pedestrian projection map orientation; pedestrian segmentation; uncalibrated monocular videos; video surveillance; Conditional probability of foregroundness; pedestrian projection map; pedestrian segmentation;
Journal_Title :
Signal Processing Letters, IEEE
DOI :
10.1109/LSP.2009.2018318