2006 Volume 16 Issue 1 Pages 1_15-1_27
This paper proposes a method for Japanese finger character recognition using 3-dimensional (3D) scanner. Although methods for gesture or finger character recognition without wearable devices, like data gloves, are convenient and stressless, Japanese finger characters involve characters which have similar shapes and cannot be distinguished by them, because most of them recognizes the characters by 2-dimensional images. The proposed method uses distance images obtained by using 3D scanner and principal component analysis in order to recognize the input character accurately and efficiently. Experimental results with 10 testees have showed that the proposed method recognized finger characters within about 10 milliseconds, a significantly improvement in comparison with an existing method, while the recognition accuracy was kept almost equivalent to the existing method.
