Bidirectional Matching Algorithm for Target Tracking Based on SIFT - Advanced Research on Computer Education, Simulation and Modeling - page 258

From the result of the experiment we can see that, it removes the wrong matching
points effectively under the bidirectional matching algorithm. There are eighteen
frames could not find their best frame in the totally eighty frames under the original
matching way ,the ratio of right matching is 78% by statistics, but when using the
bidirectional matching algorithm, there are only seven frames fail to find their best
template, the ratio of right matching is 92%.
5 Conclusions
The article uses the bidirectional matching algorithm that under SIFT to track the target.
The result of the experiment shows that the match is more stable when using the
bidirectional matching algorithm than original matching way, and it have a good
quantity of the matching points, so it has important significant in actual world using.
But it takes too much time from detecting to find the best template for each frame, it can
not meet real-time need. We can find a way to update the template automatic, then the
time of template selecting will be reduced, and it can meet the real-time need.
References
1. Cao, X., Wang, W.: Improved Image Matching Based on SIFT Algorithm. Department of
Information Science and Engineering, Shanghai Maritime University 200135
2. Ji, H., Wu, Y., Sun, H., Wang, Y.: SIFT feature matching algorithm with global information.
Optics and Precision Engineering 17(2) (2009)
3. Lowe, D.G.: Distinctive Image Features from Scale-Invariant Keypoints, January 5 (2004)
4. Chen, Z.: The SIFT Research and Implementation Based on the Image Registration (2008)
5. Zhang, S., Song, H., Xiang, X., Zhao, Y.: Fast SIFT Algorithm for Object Recognition.
Computer Systems & Applications (06) (2010)