Yang-Keun Ahn, Kwang-Soon Choi, Young-Choong Park
Gesture, Hand Gesture, Gesture Recognition, Text Input System, $1 Recognizer, Finger Detection, V-Touch
This paper presents an approach to input Korean letters by using depth information originating from a single depth camera. An index finger is detected based on the hand area of a user, which is extracted from the depth information. This paper suggests a way of recognizing gestures generated by tracking information on the index fingertips. To this end, two main approaches for recognition are suggested. The first is to simplify a gesture. A gesture is simplified with the use of specific angles that are extracted by using angles between points extracted from the index finger. Based on this approach, the error rate is found to be lower than that of the case where the approach was not applied. The second approach presents a modified $1 Recognizer. The existing $1 Recognizer is too robust to angles and, accordingly, a limited number of pattern numbers can be input and often this causes confusion depending on the input pattern. In order to resolve this problem, this paper presents an approach of comparing angles between the starting point and the end point from the center of the pattern. An experiment showed improved performance based on the approach.
Cite this paper
Yang-Keun Ahn, Kwang-Soon Choi, Young-Choong Park. (2016) The Text Input System Using Spatial Fingertip Tracking Method. International Journal of Circuits and Electronics, 1, 151-156