#### DMCA

## Nonparametric model for background subtraction (2000)

Venue: | in ECCV ’00 |

Citations: | 547 - 17 self |

### Citations

1482 | Pfinder: Real-Time Tracking of the Human Body",
- Wren, Azarbayejani, et al.
- 1997
(Show Context)
Citation Context ...This basic Normal model can adapt to slow changes in the scene (for example, illumination changes) by recursively updating the model using a simple adaptive lter. This basic adaptive model is used in =-=[1]-=-, also Kalman ltering for adaptation is used in [2{4]. In many visual surveillance applications that work with outdoor scenes, the background of the scene contains many non-static objects such as tree... |

256 | Image Segmentation in Video Sequences: A Probabilistic Approach",
- Friedman, Russell
- 1997
(Show Context)
Citation Context ... Figure 3-a shows the intensity histogram for this pixel. It is clearsthat intensity distribution is multi-modal so that the Normal distribution model for the pixel intensity/color would not hold. In =-=[5]-=- a mixture of three Normal distributions was used to model the pixel value for tra c surveillance applications. The pixel intensity was modeled as aweighted mixture of three Normal distributions: road... |

247 | Using adaptive tracking to classify and monitor activities in a site, Computer Vision and Pattern Recognition,
- Grimson, Stauffer, et al.
- 1998
(Show Context)
Citation Context ...he model. Although, in this case, the pixel intensity is modeled with three distributions, still the uni-modal distribution assumption is used for the scene background, i.e. the road distribution. In =-=[6, 7]-=- a generalization to the previous approach was presented. The pixel intensity ismodeledby a mixture of K Gaussian distributions (K is a small number from 3 to 5) to model variations in the background ... |

80 |
Computer Image Processing and Recognition ,
- Hall
- 1979
(Show Context)
Citation Context ... coordinates helps suppressing shadows, they have the disadvantage of losing lightness information. Lightness is related to the di erence in whiteness, blackness and grayness between di erent objects =-=[10]-=-. For example, consider the case where the target wears a white shirt and walks against a gray background. In this case there is no color information. Since both white and gray havethe same chromatici... |

62 |
Moving object recognition using and adaptive background memory,”in Time-Varying Image Processing and Moving Object Recognition,
- Karmann, Brandt
- 1990
(Show Context)
Citation Context ...the detection result as an update decision. The problem with this approach is that any incorrect detection decision will result in persistent incorrect detection later, which is a deadlock situations =-=[2]-=-. So for example, if a tree branch might be displaced and stayed xed in the new location for a long time, it would be continually detected.sThe second approach does not su er from this deadlock situat... |

18 |
Multivariate Density Estimation. A Wiley-Interscience Publication.
- Scott
- 1992
(Show Context)
Citation Context ...� xN be a recent sample of intensity values for a pixel. Using this sample, the probability density function that this pixel will have intensity value xt at time t can be non-parametrically estimated =-=[8]-=- using the kernel estimatorK as Pr(xt) = 1 n NX i=1 K(xt ; xi) (3) If we choose our kernel estimator function, K, to be a Normal function N(0� ), where represents the kernel function bandwidth, then t... |

6 |
Towards robust automatic tra scene analysis in real-time
- Koller, Weber, et al.
- 1994
(Show Context)
Citation Context ...tor function, K, to be a Normal function N(0� ), where represents the kernel function bandwidth, then the density canbe estimated as Pr(xt) = 1 N NX i=1 1 (2 ) d 2 j j 1 2 e ; 1 2 (xt;xi)T ;1 (xt;xi) =-=(4)-=- If we assume independence between the di erent colorchannels with a di erent 2 kernel bandwidths j for the jth color channel, then and the density estimation is reduced to Pr(xt) = 1 N = NX 0 @ 2 1 0... |

4 |
Moving object segmentation based on adabtive reference images
- Karmann, Brandt, et al.
- 1990
(Show Context)
Citation Context ...s sample, the probability density function that this pixel will have intensity value xt at time t can be non-parametrically estimated [8] using the kernel estimatorK as Pr(xt) = 1 n NX i=1 K(xt ; xi) =-=(3)-=- If we choose our kernel estimator function, K, to be a Normal function N(0� ), where represents the kernel function bandwidth, then the density canbe estimated as Pr(xt) = 1 N NX i=1 1 (2 ) d 2 j j 1... |

1 |
C.Stau er, \Adaptive background mixture models for realtime tracking
- Grimson
- 1999
(Show Context)
Citation Context ...he model. Although, in this case, the pixel intensity is modeled with three distributions, still the uni-modal distribution assumption is used for the scene background, i.e. the road distribution. In =-=[6, 7]-=- a generalization to the previous approach was presented. The pixel intensity ismodeledby a mixture of K Gaussian distributions (K is a small number from 3 to 5) to model variations in the background ... |

1 |
S.Russell, \Towards robust automatic traÆc scene analyis in real-time
- Koller, Weber, et al.
- 1994
(Show Context)
Citation Context ...n the scene (for example, illumination changes) by recursively updating the model using a simple adaptiveslter. This basic adaptive model is used in [1], also Kalmansltering for adaptation is used in =-=[2, 3, 4]-=-. In many visual surveillance applications that work with outdoor scenes, the background of the scene contains many non-static objects such as tree branches and bushes whose movement depends on the wi... |