positionit android app

24
INDEX 1.Why mobile media? 2.Special characteristics and requirements. 3.Demands for remote target positioning. 4. Positionit 5. Major components of positionit. i. Single-image-based remote target localization. ii. Two-image-based remote target localization. iii. Video-based remote-moving-target tracking. 6. Performance. 7. Challenges : Mobile media . 8. Conclusion. 9.References.

Upload: sherin-green

Post on 12-Apr-2017

161 views

Category:

Software


0 download

TRANSCRIPT

Page 1: Positionit android app

INDEX1.Why mobile media?

2.Special characteristics and requirements.

3.Demands for remote target positioning.

4. Positionit 5. Major components of positionit.i. Single-image-based remote target localization.ii. Two-image-based remote target localization.iii. Video-based remote-moving-target tracking.

6. Performance.

7. Challenges : Mobile media .

8. Conclusion.

9.References.

Page 2: Positionit android app

Remote target-localization and tracking system uses networked smartphones to localize a remote target of interest.

System can switch to both standalone and network

mode.

To showcase the power of mobile media, we present PositionIt, a mobile media-based remote target-localization and tracking system.

INTRODUCTION

Page 3: Positionit android app

Sensors in smartphones have enhanced mobile technology.

GPS -> measures phone’s positionDigital Compass-> direction of the remote targetDigital Cameras -> used for distance estimation.

High-resolution camera increased multimedia research opportunities.

WHY MOBILE MEDIA?

Page 4: Positionit android app

Lightweight computing: Optimized usage of battery power.

Performance tradeoffs : critical to achieve a good balance between accuracy and complexity.

Interactive user interfaces : Smartphones provide an interactive and friendly user interface that can be leveraged to improve performance.

SPECIAL CHARACTERISTICS AND REQUIREMENTS

Page 5: Positionit android app

Military and commercial applicationsExamples : 1. On the battlefield, soldiers can identify

colleague.2. A golfer who wants to estimate the distance to

the hole or 3. A hunter who needs to know the distance to a

target.4. A tourist might wish to associate a remote

object’s GPS location (using Google Maps or Google Earth, for example) when taking a picture so as to share the local information with others.

DEMANDS FOR REMOTE TARGET POSITIONING

Page 6: Positionit android app

Recently developed a image- and video-based remote target localization and tracking system on Android smartphones.

PositionIt - uses networked smartphones to localize a remote target by fusing data sensed by embedded sensors.

Important PositionIt feature - Computing is done on the phone and thus runs offline.

POSITIONIT

Page 7: Positionit android app

Compass reading and GPS coordinates are collected to track a remote object, by the PositionIt System.

The user then captures an image or video of the remote object to compute its distance/position.

System’s three main functional components: o Single-image based localizationo Two-image based localization,o Video-based remote moving target

tracking.

POSITIONIT (cont…)

Page 8: Positionit android app

POSITIONIT

Page 9: Positionit android app

MAJOR COMPONENTS OF POSITIONIT

Page 10: Positionit android app

Locate remote objects with regular shapes.

the user is expected to know the object’s physical height and width.

This is a camera-pose-estimation process.

The user’s input of the tight bounding box is an important step for accurate distance estimation.

An object moving on a level surface the physical height information alone is sufficient for object-position estimation.

SINGLE-IMAGE-BASED REMOTETARGET LOCALIZATION

Page 11: Positionit android app

SINGLE IMAGE BASED

Page 12: Positionit android app

PositionIt can estimate its position given two images of the remote target.

Two different phones or with one phone from two different positions.

if only the remote target’s distance is of interest, then it is sufficient to estimate the distance between the two phones,

TWO-IMAGE-BASED REMOTE TARGET LOCALIZATION

Page 13: Positionit android app

When the user collects the two images, he can identify the remote target by inputting a rectangular box enclosing the target on both photos.

After which the application will perform image feature extraction and matching, camera relative position estimation, and remote target-position estimation based on two-view triangulation.

If the two images are taken on different phones by two different users, PositionIt utilizes Android phone’s Bluetooth or WiFi net- work capabilities to send images from one device to the other.

TWO-IMAGE-BASED (cont…)

Page 14: Positionit android app

TWO IMAGE BASED REMOTE TARGET LOCALIZATION

Page 15: Positionit android app

The moving-target tracking function is available if the remote target’s physical size is known.

Takes a short video clip of the moving object and then simply draws a rough rectangular (with a finger touch and hold)

The system then derives a tight bounding box around the target object in all video frames.

With the help of the rectangle drawn by the user, the system can achieve better accuracy and higher efficiency when locating the tight bounding box.

VIDEO-BASED REMOTE-MOVING-TARGET TRACKING

Page 16: Positionit android app

The system then projects the moving object’s position on the image plane to the real-world coordinates.

While the video is being taken, the smart- phone’s GPS coordinates and digital compass readings are recorded so that the moving object’s trajectory can be drawn on the map.

VIDEO-BASED (cont…)

Page 17: Positionit android app

VIDEO-BASED REMOTE-MOVING-TARGET TRACKING

Page 18: Positionit android app
Page 19: Positionit android app

Average accuracy of:Single- image-based = 90% Two-image-based 83%

Limitations of two image based◦ 1. Distance between the two cameras is an

estimated value based on the single-image-based system, so errors will propagate during the two-image-based estimation.

◦ 2. If the object is far away and its image is small, the feature detection and matching accuracy becomes a constraint.

PERFORMANCE

Page 20: Positionit android app

The video-based tight-bounding-box extraction is generally robust, so the video-based moving- target tracking accuracy is close to the single- image-based system.

PERFORMANCE(cont…)

Page 21: Positionit android app

accuracy of some embedded sensors such as the GPS and digital compass is still limited.

Speed wise◦ the single-image-based localization can generate the

result instantly after the tight bounding box is provided.

◦ For two- image-based localization, the application generates the results within approximately 25 seconds after some initial algorithm optimization.

◦ The video-based moving-target tracking runs at approximately 5 to 7 frames per second (fps) after some initial algorithm optimization.

CHALLENGES : MOBILE MEDIA

Page 22: Positionit android app

Multimedia computing is a computationally intensive task, but mobile devices are limited by battery and computing power.

Multimedia computing could benefit from mobile devices’ all-in-one embedded sensors

Mobile devices user-friendly interfaces can help multimedia processing by lowering the complexity and improving estimation accuracy.

Mobile multimedia systems will greatly contribute to both commercial and military applications.

CONCLUSION

Page 23: Positionit android app

REFERENCES1. H. Hile et al., ‘‘Landmark-Based Pedestrian Navigation with

Enhanced Spatial Reasoning,’’ Proc.7th Int’l Conf. Pervasive Computing, Springer, 2009, pp. 5976.

2. F.X. Yu, R. Ji, and S.-F. Chang, ‘‘Active Query Sensing for Mobile Location Search,’’ Proc. ACMInt’l Conf. Multimedia (ACM MM), ACM Press, 2011, pp. 312. 

3.  J. Roters, X. Jiang, and K. Rothaus, ‘‘Recognition of Traffic Lights in Live Video Streams on Mobile Devices,’’ IEEE Trans. Circuits and Systems for Video Technology, vol. 21, no. 10, 2011, pp. 14971511.

4.  Q. Wang et al., ‘‘PositionIt an Image-Based Remote Target Localization System on Smartphones,’’ Proc. ACM Int’l Multimedia Conf., ACM Press, 2011, pp. 821822.

5.  Q. Wang et al., ‘‘Video Based Real-World Remote Target Tracking on Smartphones,’’ to be published in Proc. IEEE Int’l Conf. Multimedia and Expo, IEEE CS Press, 2012.

Page 24: Positionit android app

THANK YOU