Joint Probabilistic Data Association Filter for Real- Time Multiple Human Tracking in Video
نویسندگان
چکیده
Human tracking in video is required for interactive multimedia, action recognition, and surveillance. Two of the main challenges in tracking are modelling adequately features for tracking and resolving data (measurement) ambiguities in order to map out trajectories. A silhouette based tracker with reduced complexity joint probabilistic data association filter for resolution of measurement-to-track association problems is presented. The three main contributions of the paper are, firstly, the use of geometric constraints in reducing measurements uncertainty to any arbitrary level of accuracy. Secondly, the tracker is implemented as table lookup enabling parallel tracking of multiple objects. And thirdly to track individuals in a group, multiple feature clustering is applied to the silhouette region to identify homogeneous region for tracking. Tracking as a temporal filter is used to remove spurious motion patterns. Result based on four video test sequences demonstrates real-time performance and high accuracy over mean shift tracking given the same initial human locations as input. Keywords— Joint probabilistic data association filter, Kalman prediction, human tracking
منابع مشابه
Multiple Target Tracking With a 2-D Radar Using the JPDAF Algorithm and Combined Motion Model
Multiple target tracking (MTT) is taken into account as one of the most important topics in tracking targets with radars. In this paper, the MTT problem is used for estimating the position of multiple targets when a 2-D radar is employed to gather measurements. To do so, the Joint Probabilistic Data Association Filter (JPDAF) approach is applied to tracking the position of multiple targets. To ...
متن کاملHuman Tracking Using Improved Sample-Based Joint Probabilistic Data Association Filter
The human tracking problem is a hot issue in human-robot interaction, in which a conventional algorithm sample-based joint probabilistic data association filters (SJPDAF) is widely used. In this paper, the algorithm is first extended to the situation of multi-sensor fusion and then accelerated to promote the real-time performance. The simulation and experiments on robots both show good results,...
متن کاملTracking Multiple Speakers with Probabilistic Data Association Filters
In prior work, we developed a speaker tracking system based on an extended Kalman filter using time delays of arrival (TDOAs) as acoustic features. In particular, the TDOAs comprised the observation associated with an iterated extended Kalman filter (IEKF) whose state corresponds to the speaker position. In other work, we followed the same approach to develop a system that could use both audio ...
متن کاملJPDAF Based HMM or Real-Time Contour Tracking
Tracking objects using multiple cues yields more robust results. The well-known hidden Markov model (HMM) provides a powerful framework to incorporate multiple cues by expanding its observation. However, a plain HMM does not capture the inter-correlation between measurements of neighboring states when computing the transition probabilities. This can seriously damage the tracking performance. To...
متن کاملA Data Association Algorithm for Multiple Object Tracking in Video Sequences
This paper presents a particle filtering algorithm for multiple object tracking. The proposed particle filter (PF) embeds a data association technique based on the joint probabilistic data association (JPDA) which handles the uncertainty of the measurement origin.
متن کامل