Background tracking of a video taken from a front camera of non maneuvering vehicle

Download

index.pdf

Date

2014

Author

Ünver, Önder

Metadata

Show full item record

Item Usage Stats

199
views

97
downloads

In this study, a novel background tracking technique is proposed that uses extended Kalman Gaussian mixture probability hypothesis density filtering approach. Since the background in a movie, taken from a front camera of a non maneuvering moving vehicle, exhibits a non-stationary nature, tracking the background is usually done by using pixel-wise comparisons in consequent frames. Besides, some methods use features of the background to track it. The proposed method uses the feature tracking approach. The features are chosen as the corner points extracted from each video frame by using Harris corner detector. Linear motion model and non-linear measurement model are developed to predict and update the states of the features. Based on these models, the time varying number of features are tracked by extended Kalman Gaussian mixture probability hypothesis density filter. The method propagates the intensities of the targets based on random set theory and the Kalman filtering approach. MATLAB environment is used to implement the proposed background tracking method. Some simulated results of proposed method are shown for different conditions. The results indicate that the proposed method can be used for background tracking of a video instead of classical background tracking methods under some assumptions.

Subject Keywords

Image processing., Computer vision., Tracking (Engineering)., Kalman filtering., Gaussian distribution.

URI

http://etd.lib.metu.edu.tr/upload/12617030/index.pdf
https://hdl.handle.net/11511/23445

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Multi-target tracking using passive doppler measurements Guldogan, Mehmet B.; Orguner, Umut; Gustafsson, Fredrik (2013-04-26) In this paper, we analyze the performance of the Gaussian mixture probability hypothesis density (GM-PHD) filter in tracking multiple non-cooperative targets using Doppler-only measurements in a passive sensor network. Clutter, missed detections and multi-static Doppler variances are incorporated into a realistic multi-target scenario. Simulation results show that the GM-PHD filter successfully tracks multiple targets using only Doppler shift measurements in a passive multi-static scenario.
Interacting multiple model probabilistic data association filter using random matrices for extended target tracking Özpak, Ezgi; Orguner, Umut; Department of Electrical and Electronics Engineering (2018) In this thesis, an Interacting Multiple Model – Probabilistic Data Association (IMM-PDA) filter for tracking extended targets using random matrices is proposed. Unlike the extended target trackers in the literature which use multiple alternative partitionings/clusterings of the set of measurements, the algorithm proposed here considers a single partitioning/clustering of the measurement data which makes it suitable for applications with low computational resources. When the IMM-PDA filter uses clustered mea...
Gaussian mixture PHD filter for multi-target tracking using passive doppler-only measurements Guldogan, Mehmet B.; Orguner, Umut; Gustafsson, Fredrik (2012-05-17) In this paper, we analyze the performance of the Gaussian mixture probability hypothesis density (GM-PHD) filter in tracking multiple non-cooperative targets using a passive sensor network. Non-cooperative transmissions from illuminators of opportunity like GSM base stations, FM radio transmitters or digital broadcasters are exploited by non-directional separately located Doppler measuring sensors. Clutter, missed detections and multi-static Doppler variances are incorporated into a realistic multi-target s...
Use of probability hypothesis density filter for human activity recognition Günay, Elif Erdem; Akar, Gözde; Demirekler, Mübeccel; Department of Electrical and Electronics Engineering (2016) This thesis addresses a Gaussian Mixture Probability Hypothesis Density (GMPHD) based probabilistic group tracking approach to human action recognition problem. First of all, feature set of the video images denoted as observations are obtained by applying Harris Corner Detector(HCD) technique following a GMPHD lter, which is a state-of-the-art target tracking method. Discriminative information is extracted from the output of the GM-PHD lter and using these, recognition features are constructed related to ...
Comparison of Kalman filter and Bayesian-MAP approaches in the spatio-temporal solution of the inverse electrocardiography Ters elektrokardiyografinin zaman-uzamsal çözümünde Kalman filtre ve Bayes-MAP yöntemlerinin karşilaştirilmasi Aydin, Ümit; Serinağaoğlu Doğrusöz, Yeşim (2010-07-15) In this study some of the spatial and spatio-temporal methods for the solution of the inverse problem of electrocardiography (ECG) are compared with each other. Comparisons are also made for the cases with geometric errors, where the location of the heart is shifted for 10mm and the size of the heart is reduced by 5%. The compared methods are the Kalman filter and Bayesian maximum a posteriori estimation (MAP). Two different Bayesian-MAP algorithms are used. While one uses only spatial information the other...

Citation Formats

Ö. Ünver, “Background tracking of a video taken from a front camera of non maneuvering vehicle,” M.S. - Master of Science, Middle East Technical University, 2014.