Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Background subtraction based on phase feature and distance transform
Xue G., Sun J., Song L. Pattern Recognition Letters33 (12):1601-1613,2012.Type:Article
Date Reviewed: Jun 7 2013

Xue et al. present a new background subtraction algorithm for images with various degrees of complexity in this detailed paper. The high-level algorithm is presented in figure 7. At its core is a new phase-based model created for background representation that uses a Gabor filter bank and blob aggregation. The main purpose of the research is to extract (and track) a moving object of interest from a background in diverse indoor/outdoor lighting conditions.

The authors evaluate their results visually with concrete examples. The paper is very well illustrated and supplemented with examples, results, and graphs, making it easier to quickly grasp the ideas and evaluate the results. (There are 18 figures overall, including a number of sub-figure graphs, processed camera stills, algorithms, and three tables with results.)

Sufficient theoretical and mathematical background for the approach is provided, in order to validate its soundness. The authors also produce standard precision/recall metrics from comparisons of their algorithm with known others, such as the Gaussian mixture model (GMM), kernel density estimation (KDE), and local binary pattern (LBP). The proposed algorithm achieved higher precision than those algorithms, and showed fair recall in two cases of wavering curtains and rippling water and relatively poor recall in the restaurant and shopping center environments. However, it performed much better under the last two light switch (illumination) conditions (summarized in table 1). The authors conducted other comparisons and experiments, subsequently tuning some of the parameters for specific test sequences. They discuss the limitations and complexity of their approach in section 6.3, including the cases mentioned and runtime performance that is orders of magnitude slower than some of the comparison algorithms.

Overall, this is certainly a useful result--practically and theoretically--for applications of motion tracking in dynamic environments. However, the authors will need to do more work to optimize the runtime performance aspect.

Reviewer:  Serguei A. Mokhov Review #: CR141267 (1308-0744)
Bookmark and Share
  Featured Reviewer  
 
Computer Vision (I.5.4 ... )
 
 
Object Recognition (I.4.8 ... )
 
Would you recommend this review?
yes
no
Other reviews under "Computer Vision": Date
Machine vision
Vernon D., Prentice-Hall, Inc., Upper Saddle River, NJ, 1991. Type: Book (9780135433980)
Oct 1 1992
The perception of multiple objects
Mozer M., MIT Press, Cambridge, MA, 1991. Type: Book (9780262132701)
Mar 1 1993
Computer vision, models and inspection
Marshall A., Martin R., World Scientific Publishing Co., Inc., River Edge, NJ, 1992. Type: Book (9789810207724)
Jun 1 1993
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy