Application of the Affine Transform Invariant Model to Cell Tracking

In target tracking, the movement of the target may involve translation as well as rotation, and the scale of the target in the image may also vary between frames. Severe image clutter, which may mimic the image intensity features of the target and distract the tracker, is another problem. In this paper, we develop a tracking method that is robust in the presence of image clutter and is able to accommodate linear affine transformations (translation, rotation, and/or scaling) of the target without the necessity of determining the exact transformations and the constituent parameters. Based on the affine transform invariant property, an energy function is defined such that locating the target contour in subsequent frames is equivalent to the minimization of the energy function. The particle filter is utilized to estimate the optimal solution