Estimation of motion from sequential images using integral constraints

摘要

A novel method for estimating nonuniform motion from sequential images is presented which is based on integral brightness constancy constraints. The equations which transform arbitrary integrals between sequential images are expressed as a function of the displacement field and intensity changes. The apparent displacement field is then computed by iterative projections onto the solution space of each linearized transformation equation. Some properties of this technique are demonstrated on a synthetic image pair. Realistic motion estimation is also demonstrated using sequential tagged cine magnetic resonance images.