Enhanced foreground segmentation and tracking combining Bayesian background, shadow and foreground modeling

Gallego, Jaime; Pardís, Montse; Haro, Gloria

doi:10.1016/j.patrec.2012.05.004

Cited by 16 publications

(12 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Foreground segmentation: Foreground segmentation is computed by means of the system proposed in [6], thus combining in a Bayesian MRF-MAP framework pixel-wise background model with SCGMM and SCGM foreground and shadow models respectively.…”

Section: Proposed Methodsmentioning

confidence: 99%

“…In order to achieve it, we use the planar foreground segmentation system proposed in [6] in each one of the views. Once the foreground object has been initialized and segmented in all the views, we use conservative Visual Hull reconstruction with tolerance τ = 1, in order to achieve the voxelized 3D volume.…”

Section: Initializationmentioning

confidence: 99%

“…We estimate how many Gaussians are needed for correctly modeling the object analogously to the proposal presented in [6], i.e. by analyzing the color histogram for this purpose.…”

Section: Initializationmentioning

confidence: 99%

“…Hence, we maintain the consistency of the foreground model, in order to give robustness to the overall system. Also, in order to achieve a better adaptation of the model into the silhouette of the object, we apply a Gaussian split criterion presented in [6], according to the spatial size of the Gaussian. Gaussians with big area are split into two smaller Gaussians in the direction of the eigenvector associated to the largest eigenvalue (λmax).…”

Section: Regularization Of the Gaussians Displacementsmentioning

confidence: 99%

“…Therefore, for each one of the views Cj, we project the visible Gaussians belonging to ζ C j according to the projection matrices and focal length that each camera sensor presents. These Gaussians will be used in the 2D planar foreground segmentation for each camera according to [6].…”

Section: Projecting 3d Foreground Model To 2dmentioning

confidence: 99%

See 4 more Smart Citations

Multiview foreground segmentation using 3D probabilistic model

Gallego

Pardàs

2014

2014 IEEE International Conference on Image Processing (ICIP)

Self Cite

View full text Add to dashboard Cite

We propose a complete multi-view foreground segmentation and 3D reconstruction system that defines a 3-dimensional probabilistic model to model the foreground object in the 3 spatial dimensions, thus gathering the information from all the camera views. This 3D model is projected to each one of the views in order to perform the 2D segmentation with the foreground information shared by all the cameras. Then, for each one of the views, a MAP-MRF classification framework is applied between the projected region-based foreground model, the pixel-wise background model and the regionbased shadow model defined for each view. The resultant masks are used to compute the next 3-dimensional reconstruction. This system achieves correct results by reducing the false positive and false negative errors in sequences where some camera sensors can present camouflage situations between foreground and background. Moreover, the use of the 3D model opens possibilities to use it for objects recognition or human activity understanding.Index Terms-3D reconstruction, Multi-view foreground segmentation, 3D probabilistic model, SCGMM.

show abstract

Section: Proposed Methodsmentioning

confidence: 99%

Section: Initializationmentioning

confidence: 99%

“…We estimate how many Gaussians are needed for correctly modeling the object analogously to the proposal presented in [6], i.e. by analyzing the color histogram for this purpose.…”

Section: Initializationmentioning

confidence: 99%

Section: Regularization Of the Gaussians Displacementsmentioning

confidence: 99%

Section: Projecting 3d Foreground Model To 2dmentioning

confidence: 99%

See 3 more Smart Citations