Modular Tracking Framework: A Unified Approach to Registration based Tracking

This paper presents a modular, extensible and highly efficient open source framework for registration based tracking. It is implemented entirely in C++ and is designed from the ground up to easily integrate with systems that support any of several major vision and robotics libraries including OpenCV, ROS and Eigen. To establish the theoretical basis for the design of this system, we introduce a new way to study registration based trackers by decomposing them into three constituent sub modules while also extending the unifying formulation described in \cite{Baker04lucasKanade_paper} to account for several important advances in the field since its publication. In addition to being a practical solution for fast and high precision tracking, this system can also serve as a useful research tool by allowing existing and new methods for any of the aforementioned sub modules to be studied better. When a new method for one of these sub modules is introduced in literature, this breakdown can help to experimentally find the combination of methods for the other sub modules that is optimum for it while also allowing more comprehensive comparisons with existing methods to understand its contributions better. By extensive use of generic programming, the system makes it easy to plug in a new method for any of the sub modules so that it can not only be tested with existing methods for other sub modules but also become immediately available for deployment in any system that uses the framework.