Applying multi layer homography for multi camera person tracking

Arsic, D.; Hristov, Emil; Lehment, Nicolas H.; Hornler, Benedikt; Schuller, Björn; Rigoll, Gerhard

doi:10.1109/icdsc.2008.4635731

Cited by 34 publications

(21 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…See Table 1 for an overview that highlights the way person localization and track assignment is performed -our primary paper scope. These approaches can thereafter be embedded in a state estimation framework, either recursive (Kalman [1][8] [12], particle filtering) or in batch mode (Viterbi-style MAP estimation [6], graph-cut space-time segmentation [9] or otherwise [4,10]). …”

Section: Previous Workmentioning

confidence: 99%

Multi-person Localization and Track Assignment in Overlapping Camera Views

Liem

Gavrila

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. The assignment of multiple person tracks to a set of candidate person locations in overlapping camera views is potentially computationaly intractable, as observables might depend upon visibility order, and thus upon the decision which of the candidate locations represent actual persons and which do not. In this paper, we present an approximate assignment method which consists of two stages. In a hypothesis generation stage, the similarity between track and measurement is based on a subset of observables (appearance, motion) that is independent of the classification of candidate locations. This allows the computation of the K-best assignment in low polynomial time by standard graph matching methods. In a subsequent hypothesis verification stage, the known person positions associated with the K-best solutions are used to define the full set of observables, which are used to compute the maximum likelihood assignment. We demonstrate that our method outperforms the state-of-the-art on a complex outdoor dataset.

show abstract

Section: Previous Workmentioning

confidence: 99%

Multi-person Localization and Track Assignment in Overlapping Camera Views

Liem

Gavrila

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…The advantage of the proposed approach is that the decision on correspondences across cameras is delayed until it can be performed at the object-level in a more robust manner (i.e. matching entire object appearance at once, and not requiring colour correspondences across cameras), as compared to matching individual epipolar line segments [13] or individual height layers [1,3,9]. This allows us to handle complex environments, containing a significant amount of lighting and background changes, with a moderate number of cameras, see Figure 1.…”

Section: Related Workmentioning

confidence: 99%

“…Previous work can be furthermore distinguished by the type of state estimation employed, whether recursive (Kalman [1,7,13], particle filtering) or in batch mode (Viterbi-style MAP estimation [4], graph-cut segmentation in space-time volume [9] or otherwise [2]). …”

Section: Related Workmentioning

confidence: 99%

“…Tracking can then be done based on the estimated ground plane positions. In [1] the method of using homogeneous projections of foreground blobs for detecting people is extended by using multiple projection levels. Not only are projections made on the ground plane, but multiple height levels are defined at which images are homogeneously transformed and compared, thus creating a 3D stack of 2D projections giving much more detail on the likelihood of a person's position.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Multi-person tracking with overlapping cameras in complex, dynamic environments

Liem

Gavrila

2009

Procedings of the British Machine Vision Conference 2009

View full text Add to dashboard Cite

This paper presents a multi-camera system to track multiple persons in complex, dynamic environments. Position measurements are obtained by carving out the space defined by foreground regions in the overlapping camera views and projecting these onto blobs on the ground plane. Person appearance is described in terms of the colour histograms in the various camera views of three vertical body regions (head-shoulder, torso, legs). The assignment of measurements to tracks (modelled by Kalman filters) is done in a non-greedy, global fashion based on ground plane position and colour appearance. The advantage of the proposed approach is that the decision on correspondences across cameras is delayed until it can be performed at the object-level, where it is more robust.We demonstrate the effectiveness of the proposed approach using data from three cameras overlooking a complex outdoor setting (train platform), containing a significant amount of lighting and background changes.

show abstract

“…Using the holography map, Khan and Shah [4], and Arsí c et al [5] utilized mu lti-view foreground fusion to localize pedestrians on mu ltip le parallel p lanes. Among them, Khan and Shah [4] removed some phantoms by some prior geo metry assumption, wh ile Arsí c et al [5] utilized temporary informat ion to omit the phantoms. Nevertheless, the two methods are both sensitive to some foreground informat ion in some cases.…”

Section: Related Workmentioning

confidence: 99%

Multi-camera Pedestrian Detection with Multi-view Bayesian Network Model

Peng¹,

Wang²,

Huang³

2012

Procedings of the British Machine Vision Conference 2012

View full text Add to dashboard Cite

In this paper, we propose a novel method with the multi-view Bayesian network (M BN) model to detect pedestrians from multi-camera surveillance videos. In our method, the ground plane is discretized in a predefined set of locations and our aim is to estimate the occupancy probability of each location that can be then used to predict the occurrence of pedestrians. To reduce the possible phantoms, we use M BN to model the potential occlusion relationship of all locations in all views, and the "subjective supposing" node states (SSNS) as a set of Boolean parameters of M BN to denote whether a pedestrian occurs at the corresponding location. Thus a learning algorithm is proposed to estimate the SSNS parameters, by finding such a configuration that the final occupancy possibility can best explain the image observations (i.e., foreground masks) from different views. The experimental results on the APIDIS and PETS09 S2L1 benchmark datasets show that our method can obtain at least 10% performance gain compared with several state-of-the-art algorithms.

show abstract

Applying multi layer homography for multi camera person tracking

Cited by 34 publications

References 9 publications

Multi-person Localization and Track Assignment in Overlapping Camera Views

Multi-person Localization and Track Assignment in Overlapping Camera Views

Multi-person tracking with overlapping cameras in complex, dynamic environments

Multi-camera Pedestrian Detection with Multi-view Bayesian Network Model

Contact Info

Product

Resources

About