Confident adaptive algorithms are described, evaluated,and compared with other algorithms that implement the estimationof motion. A Galerkin finite element adaptive approach isdescribed for computing optical flow, which uses an adaptive triangularmesh in which the resolution increases where motion is found tooccur. The mesh facilitates a reduction in computational effort by enablingprocessing to focus on particular objects of interest in a scene.Compared with other state-of-the-art methods in the literature ouradaptive methods show only motion where main movement is knownto occur, indicating a methodological improvement. The mesh refinement,based on detected motion, gives an alternative to methodsreported in the literature, where the adaptation is usually based on agradient intensity measure. A confidence is calculated for the detectedmotion and if this measure passes the threshold then themotion is used in the adaptive mesh refinement process. The idea ofusing the reliability hypothesis test is straightforward. The incorporationof the confidence serves the purpose of increasing the opticalflow determination reliability. Generally, the confident flow seemsmost consistent, accurate and efficient, and focuses on the mainmoving objects within the image.
|Journal||International Journal of Imaging Systems and Technology|
|Publication status||Published - 22 Sep 2006|
Bibliographical noteThis paper examines reliability measures for optical flow computation in the context of previous work – the development of adaptive grid refinement approaches – and demonstrates the incorporation of confidence measures into the computational algorithm to increase accuracy. Extensive analysis and experimentation were performed against benchmark image sequences. The work has been taken forward into technology transfer activities with funding from the Higher Education Innovation Fund to develop a vision system to analyse live video from a single camera, and funding from the EU INTERREG IIIA programme to develop a motion analysis system for hand gestures in 3D character model animation.
Reference text: P. Anandan, A computational framework and an algorithm for the measurement
of visual motion, Int J Comput Vis 2 (1989), 283–310.
A. Bab-Hadiasahar and D. Suter, Robust optic flow computation, Int J Comput
Vis 29 (1998), 59–77.
J.L. Barron, D.J. Fleet, and S.S. Beauchemin, Performance of optical flow
techniques, Int J Comput Vis 12 (1994), 43–77.
J.L. Barron and M. Khurana, Determining optical flow for large motion
using parametric models in a hierarchical framework, Proc Vis Interface,
1997, pp. 47–56.
S. Benayoun and N. Ayache, Dense non-rigid motion estimation in sequences
of medical images using differential constraints, Int J Comput Vis 26
M.J. Black and P. Anandan, The robust estimation of multiple motions:
Parametric and piecewise-smooth flow fields, Comput Vis Image Understanding
63 (1996), 75–104.
T.M. Chin, W.C. Karl, and A.S. Willsky, Probabilistic and sequential computation
of optical flow using temporal coherence, IEEE Trans Image Process
3 (1994), 773–788.
I. Cohen and I. Herlin, Non uniform multiresolution method for optical flow
and phase portrait models: Environmental applications, Int J Comput Vis 33
J.V. Condell, Motion tracking in digital images, PhD Thesis, Faculty of
Informatics, University of Ulster, 2002.
J.V. Condell, B.W. Scotney, and P.J. Morrow, Estimation of motion through
inverse finite element methods with triangular meshes, Proc Ninth Int Conf
Comput Anal Images Patterns (CAIP 2001), Warsaw, Poland. Lecture Notes
in Computer Science, Vol. 2124, Springer, Berlin, 2001, pp. 333–340.
J.V. Condell, B.W. Scotney, and P.J. Morrow, Adaptive grid refinement procedures
for efficient optical flow computation, Int J ComputVis 61 (2005),
D.J. Fleet and A.D. Jepson, Computation of normal velocity from local
phase information, Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit
(CVPR 1989), San Diego, California,1989, pp. 379–386.
L. Florack, W. Niessen, and M. Nielsen, The intrinsic structure of optic flow
incorporating measurement duality, Int J Comput Vis 27 (1998), 263–286.
B. Galvin, B. McCane, K. Novins, D. Mason, and S. Mills, Recovering
motion fields—An evaluation of eight optical flow algorithms, Proc Br Mach
Vis Conf (BMVC 1998), 1998.
A. Giachetti and V. Torre, Refinement of optical flow estimation and detection
of motion edges, Proc Fourth Eur Conf Comput Vis (ECCV 1996),
Cambridge, UK. Lecture Notes in Computer Science, Vol. 1065, Springer-
Verlag, Berlin,1996, pp. 151–160.
J.V. Graham, B.W. Scotney, and P.J. Morrow, Evaluation of inverse finite
element techniques for gradient based motion estimation, Proc Third IMA
Conf Imaging Digital Image Process, Leicester, UK, 2000, pp. 200–220.
B.K.P. Horn and B.G. Schunck, Determining optical flow, Artif Intell 17
H. Kirchner and H. Niemann, Finite element method for determination of
optical flow, Pattern Recognit Lett 13 (1992), 131–141.
S.H. Lai and B.C. Vemuri, Reliable and efficient computation of optical
flow, Int J Comput Vis 29 (1998), 87–105.
T. Lin and J.L. Barron, Image reconstruction error for optical flow, Proc Vis
Interface, 1994, pp. 73–90.
B. Lucas and T. Kanade, An iterative image registration technique with an
application to stereo vision, Proc Seventh Int Jt Conf Artif Intell (IJCAI
1981), Vancouver, British Columbia, Canada,1981, pp. 674–679.
P. Moulin, R. Krishnamurthy, and J.W. Woods, Multiscale modeling and
estimation of motion fields for video coding, IEEE Trans Image Process 6
H.H. Nagel, Displacement vectors derived from second order intensity variations
in image sequences, Comput Vis Graphics Image Process 21 (1983), 85–117.
J.M. Odobez and P. Bouthemy, Robust multiresolution estimation of parametric
motion models in complex image sequences, Proc Seventh Eur Conf
Signal Process, (EUSIPCO 1994), Edinburgh, Scotland,1994, pp. 411–415.
E.P. Ong and M. Spann, Robust optical flow computation based on least-median-
of-squares regression, Int J Comput Vis 31 (1999), 51–82.
- adaptive grids
- confidence measures
- finite element methods
- optical flow