Hand Gesture Recognition Model using Standard Deviation-based Dynamic Time Warping Technique
Jorrel J. David, John Carlo L. Genavia, Trixie L. Laplana, Lady Faith O. Rodrigo, Darwin A. Rodriguez, Roselito E. Tolentino · 2021
Gestures are common ways of human communication that represent language and emotion using signs and symbols drawn by the movements of the hands or fingers of a person. In this study, the main focus on air-written English Capital Alphabets. However, most of the systems that operate recognition of these gestures like air-drawn letters often misclassify some of them, especially those having similar slope sequence. This study focuses on the use of the Leap Motion Controller, Air-writing of English Capital Alphabets, and Dynamic Time Warping to recognize air-drawn letters. Using Dynamic Time Warping, the system can recognize letters in two-dimensional values despite the input set and the reference template having different number of points, and can also classify letters by finding the lowest difference of the standard deviation between the input and templates. The existing systems depend on the slopes or angles of the letters formed, either time-based or geometrically-based, and rely on zero difference or null set as signal of identification of a letter. The Leap Motion Controller is used to detect the position of the finger of the user, which acts as the "pen" in air-writing English Capital Alphabets. The proponents evaluated the overall reliability of the system by letting the users test the system by air-writing each of the 26 letters numerous times.