Moving Target Indication and Tracking from Moving Sensors

Pose-invariant vehicle identification in aerial electro-optical imagery

“…target detection e.g. VMTI [16] for motion imagery and saliency technique [28] for still imagery) would process aerial imagery and provide image chips (bounding-box encapsulating the target) consisting of targets within the chip. The framework was implemented in MATLAB on a 64-bit, 2.35 GHz system and in the sections that follow we provide a detailed description of the framework and its evaluation.…”

“…The target detection techniques do not provide sufficiently precise image chips encapsulating the vehicles as required by the recognition algorithm. As a result, we employ an intermediate vehicle extractor which, when cued by the VMTI [16] or saliency operator [28], segments the vehicle and fits a tightly bounding box.…”

“…This has a number of applications such as unmanned aerial vehicle (UAV) handover [18], tracking through urban/heavily occluded environments, and traffic monitoring and planning. In this study, we assume that candidate vehicles have already been detected, probably based on video moving target indication (VMTI) [16] and/or saliency detection.…”

Evaluation of a Particle Filter to Track People for Visual Surveillance

“…The particle filter follows the PJFDT methodology described in [1] and [8]. The pseudo-code for a single iteration of the algorithm is shown in Algorithm 1.…”

“…Heuristics are used to spawn new trackers or terminate unmatched ones based on association with foreground mask connected components. All algorithms were implemented in the proprietary Analysts' Detection Support System (ADSS) [8]. ADSS is a flexible processing engine that has been designed to apply a pipeline of algorithms to streaming sensor data in real-time.…”