M.Sc Thesis

M.Sc StudentKatan Hod
SubjectVisual Tracking and Characterization of Finger
DepartmentDepartment of Mechanical Engineering
Supervisor PROF. Miriam Zacksenhouse


After speech, one of the significant modes of communications among people is hand gestures. In communication with computers (Human Computer Interaction- HCI) the use of simple mechanical devices, such as a keyboard and a mouse, is common. We can divide HCI hand gesture research into two categories:
(1) Hand posture recognition or static hand gestures.
(2) Hand movement recognition or dynamic hand gestures.

In this work I developed a visual system for characterizing the movements of a single finger as part of a larger project for developing dynamic hand gesture recognition. The work focuses on the characterization of the index movement and develops two different methods for visual measurement of the joints angles. One method is based on finger kinematics model and relies on a glove with five points on it. The second method calculates the joint-angles directly by fitting straight lines to the edges of the phalanges of the finger and calculating the angle between two adjacent lines. This method does not require any external glove or objects attached to the finger.

The measurement systems were used to analyze two typical movements of the index finger, which have previously been studied by using a data-glove. Compared with the data-glove, the designed visual systems measure also the third joint and are either glove-free or involve a relatively flexible, and thus less restrictive glove. The visual measurements supports out previous characterization work, which revealed straight joint space trajectories.