Moving object segmentation based on optical flow field

Kun Zhao; Yi Zeng; Fuyuan Peng; Yan Tian; Yiping Xu

doi:10.1117/12.748475

14 November 2007 Moving object segmentation based on optical flow field

Kun Zhao, Yi Zeng, Fuyuan Peng, Yan Tian, Yiping Xu

Proceedings Volume 6789, MIPPR 2007: Medical Imaging, Parallel Processing of Images, and Optimization Techniques; 678923 (2007) https://doi.org/10.1117/12.748475
Event: International Symposium on Multispectral Image Processing and Pattern Recognition, 2007, Wuhan, China

Abstract

Traditional optical flow estimation methods have two drawbacks: firstly, flow estimation is not accurate enough on border of the target which result in the blurring there; secondly, with the increasing of the speed of the object motion, the estimation error of brightness constancy assumption will be also increased. Focusing on the above two points, an improved optical flow estimation method is presented in this paper. To alleviate flow constraint errors, we employed a re-weighted least-squares method to suppress unreliable flow constraints, thus leading to robust estimation of optical flow. In addition, a coarse-to-fine adjustment scheme is proposed to refine the optical flow estimation especially for large image motions. We also proposed an algorithm for target segmentation of image sequences based on clustering in the feature vector space. Experimental results on some synthetic and real image sequences showed that, the proposed algorithm has favorable performance comparing with the existed methods in terms of accuracy and computation cost. Furthermore, the segmentation results based on the proposed method can be obtained in the case of complicated background.

Citation Download Citation

Kun Zhao, Yi Zeng, Fuyuan Peng, Yan Tian, and Yiping Xu "Moving object segmentation based on optical flow field", Proc. SPIE 6789, MIPPR 2007: Medical Imaging, Parallel Processing of Images, and Optimization Techniques, 678923 (14 November 2007); https://doi.org/10.1117/12.748475

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available