Video Segmentation of Moving Humans for Assistive Environments

In this paper, we present two recently proposed efficient methods for human segmentation from video in indoor environments: the illumination sensitive background method and the self-organizing background subtraction (SOBS) method. Both methods maintain multiple background models. The SOBS method has been modified in this work for gray-scale frames, in order to decrease processing times. The video data are acquired indoors from a fixed fish-eye camera in the living environment. The paper presents the algorithmic implementation and modifications details, while results are also presented for a small number of video sequences.