Estimation of multiple motions using block matching and Markov random fields

Abstract

This paper deals with the problem of estimating multiple motions at points where these motions are overlaid. We present a new approach that is based on block-matching and can deal with both transparent motions and occlusions. We derive a block-matching constraint for an arbitrary number of moving layers. We use this constraint to design a hierarchical algorithm that can distinguish between the occurrence of single, transparent, and occluded motions and can thus select the appropriate local motion model. The algorithm adapts to the amount of noise in the image sequence by use of a statistical confidence test. The algorithm is further extended to deal with very noisy images by using a regularization based on Markov Random Fields. Performance is demonstrated on image sequences synthesized from natural textures with high levels of additive dynamic noise.

Multiresolution Parametric Estimation of Transparent Motions and Denoising of Fluoroscopic Images

“…In [6], dense velocity fields are computed by adding a regularization term to (2), allowing local motion variations to be correctly estimated at the price of a high sensitivity to noise and complex computations. On the contrary, strong assumptions on the velocities are made in [7] by considering v 1 and v 2 constant on blocks of the image (and therefore accounting for a limited range of motions), to allow fast and robust motion estimation. In [4], the velocities are decomposed on a B-spline basis, so that this method can account for complex motion, while staying relatively tractable.…”

Coronary Tree Extraction Using Motion Layer Separation

“…Many classical motion estimation methods have been adapted to the transparency case substituting the brightness assumption with some constraint equations. The popular methods that fall under this category are : block-matching techniques and random fields [12], regularization [13], multi-resolution [1] etc.…”

Multiresolution parametric estimation of transparent motions

“…Different methods have been proposed to estimate the transparent motions from this equation using three successive images: minimizing a global energy function with regularization [3], adapting wavelets or B-spline decomposition [4], using Markov random fields or block-matching [5]. Another way is to formulate the problem in the frequency domain [6], but the time interval over which motion has to be assumed constant is then much larger.…”