Citation Context

... robustness and efficiency of the tracker. For example, face tracking in 1 a crowded scene relies more on target representation than on target dynamics [21], while in aerial video surveillance, e.g., =-=[74]-=-, the target motion and the ego-motion of the camera are the more important components. In real-time applications only a small percentage of the system resources can be allocated for tracking, the res...

Citation Context

...e motion-planning problem for monitoring targets by MSAs . The work is motivated by the application for uninhabited air vehicles (UAVs) in both military [1]–[3], [7] and civilian surveillance systems =-=[5]-=-. The MSA here is modeled as a nonholonomic point mass moving on a two-dimensional (2-D) plane at a constant speed with a bounded turning curvature. This model is also called the Dubins car in the lit...

Citation Context

...rial Video Surveillance System The aim of the "aerial video surveillance system" (AVS) developed by the Sarnoff Corporation is to provide enemy observation, resource monitoring, and search and rescue =-=[17]-=-. The core of the AVS system is the ACADIA chip [44], for fast image alignment, moving object detection and tracking, geo-registration, and video mosaicing (i.e. stitching many frames together to crea...

Citation Context

...ay be more easily deployed in a variety of geographic areas. There has been growing research activity in the area of aerial surveillance and object tracking; examples include the excellent summary in =-=[7]-=- and several systems with dynamic object tracking, described in [8], [9]–[10], among many others. Some of these systems are similar to the work described in this paper. However, within the application...

Citation Context

...to create larger views than obtainable through the field of view of any one frame (e.g., [10, 6]). Video from UAVs can be used to create large-scale static mosaics of scenes from aerial imagery. (See =-=[7]-=- for an overview of this and related work.) [5] points out that this mosaicing of video simultaneously extends not only the spatial information but the temporal information (change capture) as well. B...

Citation Context

...rams, respectively. We conclude the chapter with some of the limitations that we intend to address in future. 2 Related Work Tracking moving objects from an aerial platform has seen numerous advances =-=[1, 3, 16, 18, 30, 31, 38]-=- in recent years. We confine our discussion to only a subset of the literature that has strong relevance with the context of this chapter. The authors of [16] present a framework that involves separat...

Citation Context

... of the video sequence. The improved performance is demonstrated on a number of colour video sequences. Index Terms— Object tracking, mean shift, Chamfer distance transform, image segmentation, Bhattacharyya coefficient 1. INTRODUCTION Real-time feature or object tracking in video sequences is an important application in computer vision, often demanding real-time operation. In general, joint spatial and temporal analysis can be used to extract and track regions of interest in the dynamic scene. Some recent exemplars include face tracking in a crowded environment [1], aerial video surveillance [2], human body tracking and behavioural analysis [3, 4]. In the latter example, in particular, only a small fraction of the computational resource can be allocated to tracking, as the rest may be required for high-level tasks such as recognition and understanding of that behaviour. Therefore, it is desirable to ensure that the tracker is as efficient as possible. The mean shift algorithm [5, 6] is a nonparametric statistical method to find the nearest mode of a point sample distribution, that has been adopted as an efficient technique for image segmentation [7] and object tracking [8]. In this p...