Multiresolution Gaussian Mixture Models for Visual Motion EstimationRoland Wilson, Andrew Calway, Multiresolution Gaussian Mixture Models for Visual Motion Estimation. Proceedings of the IEEE International Conference on Image Processing. I. Pitas, (eds.). ISBN 0-7803-6727-8, pp. 921–924. October 2001. PDF, 109 Kbytes.
This paper introduces a new generalisation of scale-space and pyramids, which combines statistical modelling with a spatial representation. The representation uses the familiar concept of multiple resolutions, but applied to a Gaussian mixture representation of the image - hence the title MGMM. It is shown that MGMM can approximate any probability density and can adapt to smooth motions. After a brief presentation of the theory, it is shown how MGMM can be applied to the estimation of visual motion.