Section: New Results

Aggregation of patch-based estimations for illumination-invariant optical flow in live cell imaging

Participants : Denis Fortun, Charles Kervrann, Patrick Bouthemy.

Paper under review.

Live cell image sequences provide a large variety of challenging situations for motion estimation. We have developed a novel optical flow estimation method in the line of work of [11] , based on a two-stage aggregation framework and designed to handle this diversity of issues. First, semi-local candidates are estimated with a combination of patch correspondences and illumination-invariant affine motion estimations. Then, one candidate is selected at each pixel in a graph-cut based global aggregation stage. This approach allows us to overcome usual limitations of existing methods such as loss of small structures with large displacements, dependency on illumination changes and over-smoothing of the motion discontinuities. The method jointly estimates the motion and illumination fields. We have compared our approach to state-of-the-art methods and have demonstrated its ability to outperform existing methods in challenging cases frequently arising in live cell imaging (see Fig. 7 ).

Partners:: Perrine Paul-Gilloteaux (UMR 144 CNRS PICT IBiSA Institut Curie)

Figure 7. Comparison of our method with the methods of [22] and [42] on a sequence of “HeLa cells” (courtesy of F. Pérez UMR 144 CNRS Institut Curie, PICT-IBiSA).

frame 1 frame 2 illumination change map                 color code map         our method method [22] method [42] vector code map