Exploiting Constraints for Effective Visual Tracking in Surveillance Applications
| Main Author: | |
|---|---|
| Language: | English |
| Published: |
The Ohio State University / OhioLINK
2012
|
| Subjects: | |
| Online Access: | http://rave.ohiolink.edu/etdc/view?acc_num=osu1331138092 |
| id |
ndltd-OhioLink-oai-etd.ohiolink.edu-osu1331138092 |
|---|---|
| record_format |
oai_dc |
| spelling |
ndltd-OhioLink-oai-etd.ohiolink.edu-osu13311380922021-08-03T06:04:48Z Exploiting Constraints for Effective Visual Tracking in Surveillance Applications Zhu, Junda Computer Engineering Electrical Engineering Visual tracking particle filtering stereo vision electronic localization structured environment <p>With the the ubiquitous deployment of surveillance cameras, hugeamounts of video data are being generated at every moment. Analyzingthe massive surveillance videos in an efficient manner has become apressing task. Visual object tracking is one of the enablingtechnologies for video analysis and has received much attention in thecomputer vision community during the last decade. Despite the recentadvances in the visual tracking research, there are still severalchallenges to the existing methods such as efficiency, accuracy,resilience to visual ambiguities, etc. To address such challenges andimprove the tracking performance, the constraints specific to thesurveillance applications need to be utilized, which have not beenthoroughly studied before. The objective of this dissertation is toexploit the constraints pertaining to the surveillance applicationsand integrate them into the probabilistic tracking framework foreffective visual tracking.</p><p>This dissertation first presents the integration of environmentconstraints into the particle filtering framework for effectivelytracking objects for the urban surveillance applications. In theseapplications, the movements of objects are constrained by structuredenvironments. Therefore, the relationship between objects andenvironments can be exploited as additional information for improvingthe performance of tracking. An environment state is introduced torepresent the relationship between the objects andenvironments. Distance transform is used to model the environmentstate. The adaptive dynamics model and environment prior are devisedfor the particle filter to fully utilize the environment informationin the tracking process.</p><p>Then the integration of electronic localization for effective visualtracking is studied. Electronic signals, like cellular, WiFi andBluetooth signals emitted from mobile phones, are ubiquitously presentand can be associated with the objects of interest. A directionalantenna is used for collecting the signals and performing roughelectronic localization. Such location information is fed into thevisual tracking algorithm as object motion constraints, so theuncertainty and search space of visual tracking are significantlyreduced.</p><p>Finally, a stereo tracking method for measuring the speed of a movingvehicle within a structured environment is presented. The stereoconstraint between the two views and the path constraint for thevehicle's motion are exploited for accurate visual tracking whichovercomes the limitation of depth accuracy in long range stereo. Inthe proposed method, visual stereo tracking and motion estimation in3D are integrated within the framework of particle filtering. Thevisual tracking processes in the two views are coupled with each othersince they are dependent upon the same 3D motion and correlated in theobservations. Considering that the vehicle's motion is physicallyconstrained by the environment, the path constraint reconstructed fromstereo views is utilized to reduce the uncertainty about the vehicle'smotion and improve the accuracy for both tracking and speed measurement.</p> 2012-06-19 English text The Ohio State University / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=osu1331138092 http://rave.ohiolink.edu/etdc/view?acc_num=osu1331138092 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws. |
| collection |
NDLTD |
| language |
English |
| sources |
NDLTD |
| topic |
Computer Engineering Electrical Engineering Visual tracking particle filtering stereo vision electronic localization structured environment |
| spellingShingle |
Computer Engineering Electrical Engineering Visual tracking particle filtering stereo vision electronic localization structured environment Zhu, Junda Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| author |
Zhu, Junda |
| author_facet |
Zhu, Junda |
| author_sort |
Zhu, Junda |
| title |
Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| title_short |
Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| title_full |
Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| title_fullStr |
Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| title_full_unstemmed |
Exploiting Constraints for Effective Visual Tracking in Surveillance Applications |
| title_sort |
exploiting constraints for effective visual tracking in surveillance applications |
| publisher |
The Ohio State University / OhioLINK |
| publishDate |
2012 |
| url |
http://rave.ohiolink.edu/etdc/view?acc_num=osu1331138092 |
| work_keys_str_mv |
AT zhujunda exploitingconstraintsforeffectivevisualtrackinginsurveillanceapplications |
| _version_ |
1719430523333378048 |