音響信号を用いたデバイス非接触型手書き文字認識手法

本研究では，音響信号を用いて，机等の平面に設置したスマートフォンの周囲で行う，指による手書き文字を認識する手法を提案する． 具体的には，スマートフォンから発信し，指に当たって反射してきた超音波と，手書き文字を書く際に生じる可聴音を利用する． 提案手法では，デバイス内蔵のマイク/スピーカを利用するため，手書き文字認識のための追加デバイスが必要ない． 提案手法の有効性を確認するための評価実験を行った結果，被験者10名に対して低雑音環境下で36種類の文字を平均77.8%の精度で認識できた． また，可聴音と超音波を組み合わせた提案手法が，一方のみを利用するよりも高い認識精度を達成することが確認できた．

Device-Free Handwritten Character Recognition Method Using Acoustic Signal

In this research, we propose a method for recognizing handwritten characters by a finger using acoustic signals. This method is carried out using a smartphone placed on a flat surface, such as a desk. Specifically, this method uses an ultrasonic wave transmitted from the smartphone, which is reflected by the finger, and an audible sound is generated when writing a handwritten character. The proposed method does not require an additional device for handwritten character recognition because it uses the microphone/speaker built into the device. Evaluation results showed that it was able to recognize 36 types of characters with an average accuracy of 77.8% in a low noise environment for 10 subjects. In addition, it was verified that combining an audible sound and an ultrasonic wave in this method achieved higher recognition accuracy than when only an audible sound or an ultrasonic wave was used.