Model Fitting for Motion Segmentation

The problem of rigid motion segmentation of trajectory data under orthography has been long solved for non-degenerate motions in the absence of noise. But because real trajectory data often incorporates noise, tracking failures, motion degeneracies and motion dependencies, recently proposed motion segmentation methods resort to non-trivial representations to achieve state of the art accuracies, at the expense of a large computational cost. This thesis proposes a method that dramatically reduces this cost (by two orders of magnitude) with minimal segmentation accuracy loss (from 98.8% achieved by the state of the art, to 97% achieved by our method on the standard Hopkins 155 dataset). Computational efficiency comes from the use of a simple but powerful motion model that explicitly incorporates mechanisms to deal with noise, outliers and motion degeneracies. Subsets of these motion models with the best balance between prediction accuracy and model complexity are efficiently ranked from a pool of candidates. Top scoring model combinations are then merged using an averaging technique to produce the final segmentation result.