Human motion segmentation and object recognition using fuzzy rules

Our goal is to develop an object recognition and motion tracking system to assist in the analysis of data for a project in the psychology and human development department. Fuzzy membership rules provide a viable solution for creating this system. We describe our development of the feature vector, rule extraction, and image segmentation. The usefulness of the system is demonstrated via an analysis of videos of human action collected as part of an on-going collaboration with researchers in the Vanderbilt Psychology and Human Development department. Results are given to show the current progress, and future goals are presented.