3-D Object Recognition and Tracking System


Description

Two novel algorithms, namely Bound Hough Transform and Error Surface Embedding , have been developed for recognizing and tracking objects with sparse range data. Given an unknown object presented into the sensor's field of view its identity from a large set of possibilities <60 objects>, and its 6 pose parameters can be determined automatically in realtime.

Using a commercial stereovision sensor and standard computing hardware, object recognition and pose determination executes at over 122 frames per second for a database of 60 objects, with a reliability of over 97%. Once an object has been recognized and its pose determined, its pose can then be tracked with even greater reliability at a rate of over 300 fps, although current sensors provide data only at ~20 fps.

Videos




Publications

[8]

Limin Shang, `Real-Time Object Recognition in Sparse Range Images Using Error Surface Embedding `, Ph.D. Thesis, Queen`s University, January 2010
[7]

Ioannou Y., Shang, L., Harrap, R., and Greenspan, M., "Local potential wellspace embedding." International Conference on 3D Digital Imaging and Modeling, 2009.
[6]


L. Shang and M. Greenspan, “Real-time Object Recognition in Sparse Range Images Using Error Surface Embedding”, International Journal of Computer Vision, Springer Netherlands, ISSN 0920-5691 (Print) 1573-1405 (Online), 1 August 2009.

[5]

Shang, L., Greenspan, M., Jasiobedzki, P., "Model-based tracking by classification in a tiny discrete pose space", IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007.

[4]

L. Shang and M. Greenspan, "Pose Determination by Potential Well Space Embedding", 3DIM 2007: The 6th International Conference on 3-D Digital Imaging and Modeling, Montreal, Quebec, Canada, Aug. 21-23, 2007, pp 297-304

[3]


Shang, L.; Jasiobedzki, P. and Greenspan, M. (2005) "Discrete pose space estimation to improve ICP-based tracking", Fifth International Conference on 3-D Digital Imaging and Modeling ( 3DIM 2005) , pp.523-530 (oral presentation).

[2]

Greenspan, M.; Shang, L. and Jasiobedzki, P. (2004) "Efficient tracking with the Bounded Hough Transform", 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’04), Volume 1, pp. 520-527. (oral presentation)

[1]

Limin Shang, `Efficient Tracking in Sparse Range Data with the Bounded Hough Transform `, Master’s Thesis, Queen`s University, August 2004.

Copyright Queen's University 2009, 2010