Fist american sign language recognition using leap motion sensor

摘要

The paper aims to solve the problem of the fist signs in American Sign Language because their recognition is not perfect due to the Euclidian distances of the fingertip to palm position used in 3D of those signs are similar, so it is difficult to recognize them. We therefore propose a system of fist American Sign Language recognition with a bare hand in depth plane by using 3D non-contact motion sensor. In this system, two patterns of the polygon area between the consecutive fingertip positions with palm position in depth plane by Shoelace formula are used to identify the fist sign language: 1) six triangles area, 2) one hexagon area, then a decision tree is applied to classify the alphabets. The results showed the 7 alphabets in fist ASL using the researcher's hand. The accuracy of the method proposed is approximately 96.1%.