Visual Tracking

Robust general tracking tools are of major interest for applications ranging from surveillance and security to robotic navigation and nanotechnology. In these applications, the objects of interest may be translated, rotated, scaled, or non-rigidly deformed, and the goal is to find trajectories and features of those objects automatically. To solve the problem of tracking, one needs to incorporate theories of image processing, segmentation (separating objects from the background), registration (object matching), control, and estimation.

The proposed tools have been successfully applied to three different categories of targets: Objects filmed by a regular hand-held or stationary camera (including people and cars); deforming objects filmed by an infrared camera; nanofluids filmed by a camera attached to a microscope (Figure 1). The project on multi-target nanofluid tracking allowed modeling of nanoplatform dynamic behavior as a function of external magnetic field. Based on this work, it may be possible to develop nanoplatforms to be utilized for the simultaneous identification, mapping, targeting and destruction of cancer cells.

Figure 1