A Library for Implementing the Multiple Hypothesis Tracking Algorithm

Abstract

The Multiple Hypothesis Tracking (MHT) algorithm is known to produce good results in difficult multi-target tracking situations. However, its implementation is not trivial, and is associated with a significant programming effort, code size and long implementation time. We propose a library which addresses these problems by providing a domain independent implementation of the most complex MHT operations. We also address the problem of applying clustering in domain independent manner.