Citation Context

...dy parts are not isolated, and may be significantly occluded. Multiple camera tracking methods often perform segmentation in each view separately, and are thus susceptible to the same problems (e.g., =-=[11, 15]-=-). Our method avoids occlusion by only tracking heads. We place a set of cameras at a high elevation, from which the heads are almost always visible. Even under these conditions, head segmentation usi...

Citation Context

...t people by using spatially distant cameras [8]. Davis et al. at the University of Maryland have developed a multi-perspective video system for human action analysis and object detection and tracking =-=[9, 10]-=-. Collins, Lipton and Kanade built a video surveillance system using multiple cameras [11]. Ozer et al. synthesized a 3D model of humans from two cameras placed approximately 90 degree apart [12]. Mat...

Citation Context

...bject tracking also bear some relevance to TOGs. Sato and Aggarwal[2] approached the problem of classifying agent interactionss, using multiple cameras and computer vision techniques. Mittal and Davis=-=[3]-=- worked with the unification of input from a wide baseline array of cameras, and used this unification for person tracking. These papers provide some background to the problem of unifying several sens...

Citation Context

...on [4] or on Bayesian Belief Networks (BBN) [5][6], and sometimes a learning phase is required [7]. Geometry-based approaches can be further subdivided into calibrated and uncalibrated approaches. In =-=[8]-=-, each camera processes the scene and obtains a set of tracks. Then, regions along the epipolar lines in pairs of cameras are matched and the midpoints of the matched segments are back-projected in 3D...

Citation Context

...el for real-time parallel media processing and built tiles of smart sensors [3, 4]. Davis et al. developed a multi-perspective video system for human action analysis and object detection and tracking =-=[5, 6]-=-. Ozer et al. synthesized a 3D model of humans from two cameras approximately perpendicular with each other [7]. The Stanford multi-camera array group uses a dense array of CMOS image sensors to captu...

Citation Context

...ut any search [5]. Applying a more weak calibration in terms of epipolar constraints or ground plane assumptions at least allows to restrict the correspondence problem to a lower dimensional subspace =-=[1, 9]-=-. In [2] and [8], respectively, calibration data is extracted in a more qualitative way by learning the relations between the fields-of-view of different cameras and transitions between them during a ...

Citation Context

...and under different illumination conditions. As a consequence, other works in the literature are based on geometrical constraints. Geometry-based approaches can be further subdivided into calibrated (=-=Mittal and Davis, 2001-=-; Yue et al., 2004) and uncalibrated (Khan and Shah, 2003) approaches. The approach proposed by (Khan and Shah, 2003) is based on the computation of the so-called Edges of Field of View (hereinafter r...

Citation Context

...dence, most “direct correspondence” approaches require the foreground region of each target be correctly extracted in each camera view. However, this requirement may not be achievable in practice. In =-=[10]-=- and [11], Mittal and Davis launched the correspondence of objects by matching the color appearance of segmented regions along epipolar lines in pairs of camera views. The mid-points of the matched re...

Citation Context

...ssification with occlusion analysis to segment people occluded in different levels of crowd density. The complexity of this algorithm is relatively high due to pixel-wise classification. Moreover, in =-=[10]-=-, Mittal and Davis describe an algorithm for detecting and tracking multiple people in a cluttered scene by using multiple synchronized cameras located far away from each other. 3.2 Occlusion Handling...