Group Detection at Camera Handoff for Collecting People Appearance in Multi-camera Systems

Abstract

Logging information on moving objects is crucial in video surveillance systems. Distributed multi-camera systems can provide the appearance of objects/people from different viewpoints and at different resolutions, allowing a more complete and precise logging of the information. This is achieved through consistent labeling to correlate collected information of the same person. This paper proposes a novel approach to consistent labeling also capable to fully characterize groups of people and to manage miss segmentations. The ground-plane homography and the epipolar geometry are automatically learned and exploited to warp objects' principal axes between overlapped cameras. A MAP estimator that exploits two contributions (forward and backward) is used to choose the most probable label configuration to be assigned at the handoff of a new object. Extensive experiments demonstrate the accuracy of the proposed method in detecting single and simultaneous handoffs, miss segmentations, and groups.

Multimedia surveillance

“…On average, about 6000 frames have been acquired simultaneously from each camera, with about three people present in the whole scene at a time (on average). Performance of the people detection and tracking module and of the consistent labeling have been deeply reported in our previous works [8,9,4]. As an example of the reliability of the system also in complex situations, Fig.…”

“…The line computed from these points must be an epipolar line since it passes through the images of the same point of image plane I i . Given at least two lines, the epipole can be located as the intersection of these lines and the fundamental matrix can be computed [4].…”

“…An example of the effectiveness of the double backward/forward probability is the full characterization of groups of people, that can be realized by exploiting the forward-backward double check. Further details and experimental results can be found in [4] …”

Consistent Labeling Approach for a PTZ Camera Based on Template Cache and Least Recently Used Replacement Strategy

“…The same object can be still marked by the same label under the cameras without overlapping FOVs. Besides, S. Calderara, et al, also proposed a consistent labeling method with high accuracy [2][3]. The method was called HECOL (Homography and Epipolar -based Consistent Labeling).…”