A Multi-transformational Model for Background Subtraction with Moving Cameras^*

Daniya Zamalieva, Alper Yilmaz, and James W. Davis

Abstract. We introduce a new approach to perform background subtraction in moving camera scenarios. Unlike previous treatments of the problem, we do not restrict the camera motion or the scene geometry. The proposed approach relies on Bayesian selection of the transformation that best describes the geometric relation between consecutive frames. Based on the selected transformation, we propagate a set of learned background and foreground appearance models using a single or a series of homography transforms. The propagated models are subjected to MAP-MRF optimization framework that combines motion, appearance, spatial, and temporal cues; the optimization process provides the final background/foreground labels. Extensive experimental evaluation with challenging videos shows that the proposed method outperforms the baseline and state-of-the-art methods in most cases.

Keywords: Background subtraction, moving camera, moving object detection

Electronic Supplementary Material:

• Electronic Supplementary Material (MP4 1,220 KB)
• Electronic Supplementary Material (MP4 1,556 KB)
• Electronic Supplementary Material (MP4 1,598 KB)
• Electronic Supplementary Material (MP4 2,087 KB)
• Electronic Supplementary Material (MP4 1,266 KB)
• Electronic Supplementary Material (MP4 2,912 KB)
• Electronic Supplementary Material (MP4 2,303 KB)
• Electronic Supplementary Material (MP4 1,463 KB)
• Electronic Supplementary Material (MP4 1,726 KB)
• Electronic Supplementary Material (MP4 1,796 KB)
• Electronic Supplementary Material (MP4 4,514 KB)
• Electronic Supplementary Material (MP4 4,658 KB)
• Electronic Supplementary Material (MP4 2,175 KB)
• Electronic Supplementary Material (MP4 2,230 KB)

LNCS 8689, p. 803 ff.

Full article in PDF | BibTeX

lncs@springer.com
© Springer International Publishing Switzerland 2014