David Tweed - Academia.edu (original) (raw)

Papers by David Tweed

This paper describes a region based approach to computingdense estimates of 2-D motion fields in ... more This paper describes a region based approach to computingdense estimates of 2-D motion fields in image sequences.By dense we mean that we would like to have anestimate of the motion at each pixel location at any pointin the sequence. In comparison with alternatives, suchas optical flow methods, region based approaches to thisproblem have a number of advantages. Since they involvecorrelating regions of pixels, they tend to be more robustin the presence of noise and, if the region window...

We present two forms of correlation for addressing the problems presented by non-translational an... more We present two forms of correlation for addressing the problems presented by non-translational and multiple mo- tions in region based motion estimation. The first -regis- tered correlation - enables affine motion parameters to be estimated; whilst the second -iterated partial correlation - enables the identification and estimation of two motion com- ponents within a region. Incorporation of both techniques in a multiresolution framework provides for more flexibility in region selection in terms of size versus robustness. We also present a local directional smoothing algorithm based on certainty measures obtained from the correlations which further improves the motion estimation. Results of experi- ments involving both types of region motions illustrate the effectiveness of the three schemes.

Image and Vision Computing, 2002

This paper describes a region based approach to computingdense estimates of 2-D motion fields in ... more This paper describes a region based approach to computingdense estimates of 2-D motion fields in image sequences.By dense we mean that we would like to have anestimate of the motion at each pixel location at any pointin the sequence. In comparison with alternatives, suchas optical flow methods, region based approaches to thisproblem have a number of advantages. Since they involvecorrelating regions of pixels, they tend to be more robustin the presence of noise and, if the region window...

We present two forms of correlation for addressing the problems presented by non-translational an... more We present two forms of correlation for addressing the problems presented by non-translational and multiple mo- tions in region based motion estimation. The first -regis- tered correlation - enables affine motion parameters to be estimated; whilst the second -iterated partial correlation - enables the identification and estimation of two motion com- ponents within a region. Incorporation of both techniques in a multiresolution framework provides for more flexibility in region selection in terms of size versus robustness. We also present a local directional smoothing algorithm based on certainty measures obtained from the correlations which further improves the motion estimation. Results of experi- ments involving both types of region motions illustrate the effectiveness of the three schemes.

Image and Vision Computing, 2002