Welcome to the InfoSci Platform
Similarity Learning for Motion Estimation

Similarity Learning for Motion Estimation

Shaohua Kevin Zhou, Jie Shao, Bogdan Georgescu, Dorin Comaniciu
Copyright: © 2009 |Pages: 22
ISBN13: 9781605661889|ISBN10: 1605661880|ISBN13 Softcover: 9781616926021|EISBN13: 9781605661896
DOI: 10.4018/978-1-60566-188-9.ch005
Cite Chapter Cite Chapter

MLA

Zhou, Shaohua Kevin, et al. "Similarity Learning for Motion Estimation." Semantic Mining Technologies for Multimedia Databases, edited by Dacheng Tao, et al., IGI Global, 2009, pp. 130-151. https://doi.org/10.4018/978-1-60566-188-9.ch005

APA

Zhou, S. K., Shao, J., Georgescu, B., & Comaniciu, D. (2009). Similarity Learning for Motion Estimation. In D. Tao, D. Xu, & X. Li (Eds.), Semantic Mining Technologies for Multimedia Databases (pp. 130-151). IGI Global. https://doi.org/10.4018/978-1-60566-188-9.ch005

Chicago

Zhou, Shaohua Kevin, et al. "Similarity Learning for Motion Estimation." In Semantic Mining Technologies for Multimedia Databases, edited by Dacheng Tao, Dong Xu, and Xuelong Li, 130-151. Hershey, PA: IGI Global, 2009. https://doi.org/10.4018/978-1-60566-188-9.ch005

Export Reference

Mendeley
Favorite

Abstract

Motion estimation necessitates an appropriate choice of similarity function. Because generic similarity functions derived from simple assumptions are insufficient to model complex yet structured appearance variations in motion estimation, the authors propose to learn a discriminative similarity function to match images under varying appearances by casting image matching into a binary classification problem. They use the LogitBoost algorithm to learn the classifier based on an annotated database that exemplifies the structured appearance variations: An image pair in correspondence is positive and an image pair out of correspondence is negative. To leverage the additional distance structure of negatives, they present a location-sensitive cascade training procedure that bootstraps negatives for later stages of the cascade from the regions closer to the positives, which enables viewing a large number of negatives and steering the training process to yield lower training and test errors. The authors apply the learned similarity function to estimating the motion for the endocardial wall of left ventricle in echocardiography and to performing visual tracking. They obtain improved performances when comparing the learned similarity function with conventional ones.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.