指腹開放型入力デバイスにおける文字ジェスチャ識別

抄録

Smartphones are indispensable devices in our daily lives. And, text input is one of the most common smartphone operations. However, it is difficult to input text while grasping an object, so a text input method that can be used under such conditions is needed. In this study, we propose a method to identify the gesture of writing with the thumb on the index finger. Fingernail device with seven photo reflectors acquires information on skin deformation at the fingertips. After extracting features from the acquired time-series data, we used a random forest to identify the features. We obtained a dataset of 26 alphabetic character gestures and calculated the average accuracy by 10-fold cross validation. Furthermore, we considered the number of required sensors and their optimal placement.