Distributed Bayesian Inference for Consistent Labeling of Tracked Objects in Nonoverlapping Camera Networks

Abstract

One of the fundamental requirements for visual surveillance using nonoverlapping camera networks is the correct labeling of tracked objects on each camera in a consistent way, in the sense that the observations of the same object at different cameras should be assigned with the same label. In this paper, we formulate this task as a Bayesian inference problem and propose a distributed inference framework in which the posterior distribution of labeling variable corresponding to each observation is calculated based solely on local information processing on each camera and mutual information exchanging between neighboring cameras. In our framework, the number of objects presenting in the monitored region does not need to be specified beforehand. Instead, it can be determined automatically on the fly. In addition, we make no assumption about the appearance distribution of a single object, but use "similarity" scores between appearance pairs as appearance likelihood for inference. To cope with the problem of missing detection, we consider an enlarged neighborhood of each camera during inference and use a mixture model to describe the higher order spatiotemporal constraints. Finally, we demonstrate the effectiveness of our method through experiments on an indoor office building dataset and an outdoor campus garden dataset.

Distributed data association in smart camera network via dual decomposition

“…(7), and the transition probability between neighboring cameras are set according to prior knowledge. Note here we use a weak but more realistic uniform model (8) instead of Gaussian model in previous works [1,12,32,43]. We set min , , 0.25…”

“…The main limitation of [12] is that it requires the knowledge of the number of objects under tracking, which is usually unavailable in practice. In a recent work [32], the authors proposed a distributed Bayesian framework in which both the sampling space and the posterior distribution of labeling variables are inferred online, based on knowledge propagation between neighboring cameras. The number of objects can be determined automatically from observation data.…”

“…where min ,  i j and max ,  i j are the minimum and maximum traveling time between camera sites i l and j l , respectively. Note that in previous works truncated Gaussian distribution was used as traveling time model [1,12,32]. Our traveling time constraint is much weaker than Gaussian, but it is more suitable when person walks in abnormal speed, for example, pauses randomly during walking.…”

Abnormal Event Detection via Multikernel Learning for Distributed Camera Networks

“…In the literature [2, 3], the framework for multiple pedestrian tracking by using overlapping cameras was presented. In [4], several major challenges in distributed video processing, including robust and computationally efficient inference and opportunistic and parsimonious sensing, were discussed.…”

Event analysis based on multiple video sensors for cooperative environment perception

“…In [8], global optical flow orientation histogram was proposed to detect abnormal vidual events. For improving the event analysis accuracy, by adopting overlapping cameras, a framework for multiple pedestrian tracking was presented in the paper [9]. The methods base on the partial information, such as in small observation windows of the image were used to analyze the events.…”