ABSTRACT
The problems of accurately tracking people in human-robot co-existing environments are considered as one of the significant problems in present times. A state-of-the-art randomized template matching algorithm, called photometric- invariant CFAsT-Match (PICFAsT-Match), can be implemented in real robot to solve an active problem within this genre, where vision sensing is used for detecting human shoes in challenging real-life environments, during pursuit. The performance of the PICFAsT-match depends on suitable choices of a parameter, named score coefficient. In this chapter, a detailed study on possible choices of the variations of the score-coefficient parameter is performed, and the overall impacts of the chosen variations on the shoe detection outcomes are evaluated.
