Navigation:

Hilfsnavigation:

Language / Sprache / Langue / Língua
zum Inhalt
zur Hauptnavigation
zur Zielgruppennavigation

Zielgruppennavigation:

Hauptnavigation:

Sprache:

Abstract
Database with GT
Videos
References

Identification of Traffic Lights for Pedestrians with Visual Impairment by the use of Mobile Devices

Master Thesis of Jan Roters

Publication

Kai Rothaus, Jan Roters, Xiaoyi Jiang:
Localization of Pedestrian Lights on Mobile Devices
Proc. of Asia Pacific Signal and Information Processing Association, Sapporo, Japan, Oct. 2009, Published

Abstract

In this work a framework is presented to localize and classify pedestrian lights using mobile devices. Our method can be applied as interactive assistance for visually handicapped people to help them passing a pedestrian crossing. Since the computation power and the storage resources of mobile devices are limited the main objective on the localization task is the efficiency of the used computer vision algorithms. The requirement on the traffic light classification is not to miss the crucial red light (reliability). We have designed a prototype for German traffic lights and realized it on a Nokia N95. The presented results demonstrate the efficiency as well as reliability of our implementation.

Image Database and Ground-Truth

Two databases are available. The first database consists of images in original resolution (480x640). It was used at the beginning of the research for traffic light detection. Since the traffic light detection approach works on low resolution video streams, a second database with low resolution images is available (240x320).
Those packages contain the Ground-Truth. In each line of the files "groundtruth_large.txt" and "groundtruth_small.txt" the following information is stored:

image number
y1
x1
y2
x2
current phase, 1 = red, 2 = green
crucial light state, 0 = other light, 1 = crucial light

To load a database using Matlab the dlmread function can be used: gt = dlmread('groundtruth_small.txt');

Video Database and Ground-Truth

The video database consists of 14 image sequences. For each sequence a file "groundtruth.txt" is available that contains Ground-Truth. To read these files refer to "Image Database and Ground-Truth"

Videos

Live - Experiment:

Download

Other videos are available for download:

Results of the 14 image sequences by use of our approach with additional video analysis and time-based verification. Sequence 01 (105 sec), 02 (98 sec), 03 (77 sec), 04 (88 sec), 05 (22 sec), 06 (34 sec), 07 (17 sec), 08 (45 sec), 09 (14 sec), 10 (60 sec), 11 (41 sec), 12 (15 sec), 13 (12 sec), 14 (51 sec),
Example sequence (30 sec): green light detection with border region control to help the pedestrian to get the entire scene into field of view
Reliable traffic light detection (18 sec): red light detection with border region control and phase switch from red to green
Traffic light detection fails (16 sec): unable to detect red lights since the colors are falsified; only a few succeeded localizations but not enough for a secure feedback

References

R. Ballagas, J. Borchers, M. Rohs, and J. G. Sheridan:
The smart phone: A ubiquitous input device.
IEEE Pervasive Computing, vol. 5, no. 1, pp. 70-77, 2006

J. Aranda and P. Mares:
Visual system to help blind people to cross the street.
ICCHP (J. Klaus, K. Miesenberger, W. L. Zagler, and D. Burger, eds.), vol. 3118 of Lecture Notes in Computer Science, pp. 454-461, Springer, 2004

V. Ivanchenko, J. Coughlan, and H. Shen:
Crosswatch: A camera phone system for orienting visually impaired pedestrians at traffic intersections.
ICCHP 2008, Linz, Austria, July 9-11, 2008. Proceedings, vol. 5105 of Lecture Notes in Computer Science, pp. 1135-1140, Springer, 2008.

J. Hannuksela, P. Sangi, and J. Heikkilä:
Vision-based motion estimation for interaction with mobile devices.
Computer Vision and Image Understanding, vol. 108, no. 1-2, pp. 188-195, 2007

B. D. Lucas and T. Kanade:
An iterative image registration technique with an application to stereo vision.
IJCAI (P. J. Hayes, ed.), pp. 674-679, William Kaufmann, 1981

A. R. Smith:
Color gamut transform pairs.
SIGGRAPH '78: Proceedings of the 5th annual conference on Computer graphics and interactive techniques , (New York, NY, USA), pp. 12-19, ACM, 1978

D. Ciresan and D. Damian:
Preserving topological information in the windowed hough transform for rectangle extraction.
DAGM-Symposium (K. Franke, K.-R. Müller, B. Nickolay, and R. Schäfer, eds.), vol. 4174 of Lecture Notes in Computer Science, pp. 172-181, Springer, 2006.

Warning: include(studieren.inc) [function.include]: failed to open stream: No such file or directory in /data/wwwdata-weberknecht/httpd-8029/htdocs/block_f.inc on line 15

Warning: include() [function.include]: Failed opening 'studieren.inc' for inclusion (include_path='.:..:../..:../../..:../../../..:../../../../..:/opt/wwwdata/httpd-8000/htdocs:.:/opt/wwwdata/common/php') in /data/wwwdata-weberknecht/httpd-8029/htdocs/block_f.inc on line 15