A reliability index that evaluates the quality of optical flow is needed because the accuracy of estimation of camera movement or the shape of the environment depends on the accuracy of optical flow. A reliability index using spatial brightness gradient information has already been proposed, and its effectiveness has been confirmed. In this paper, we propose a reliability index that considers the error margin of matches in addition to the spatial brightness gradient information. The result of experiments showed that considering the error margin of matches increases the performance of the reliability index. It was also found that the camera movement can be estimated from optical flow in a synthetic image to which the camera movement was known beforehand. Then, it was confirmed that rejecting using the reliability index improves the accuracy of estimation. Moreover, in an experiment in witch changing of noise was added to the image sequence, it was shown that moderate noise stabilized the movement estimation
[1]
David J. Fleet,et al.
Performance of optical flow techniques
,
1994,
International Journal of Computer Vision.
[2]
Takeo Kanade,et al.
An Iterative Image Registration Technique with an Application to Stereo Vision
,
1981,
IJCAI.
[3]
Edward H. Adelson,et al.
The Laplacian Pyramid as a Compact Image Code
,
1983,
IEEE Trans. Commun..
[4]
Berthold K. P. Horn,et al.
Determining Optical Flow
,
1981,
Other Conferences.
[5]
Naoya Ohta.
Image movement detection with reliability indices
,
1991
.
[6]
Kenichi Kanatani,et al.
Do We Really Have to Consider Covariance Matrices for Image Feature Points
,
2002
.