I have some incomplete and noisy point observations (x,y coordinate) of particles moving in a spatially and temporally dynamic flow in 2D space. The observations are at uneven time steps, and points may drop in or out at each time step.
Is there a technique/algorithm that would help me estimate the displacement fields between each/all time steps while optimizing some constraints (overall consistency of the displacement fields across time stamps, for example), such that I might be able to establish the highest confidence trajectories for a subset of points?
I'm at a loss for search topics...