Evaluation of finger position estimation with a small ranging sensor array
Yu Ishikawa, Buntarou Shizuki, Junichi Hoshino · 2017
We implemented the input interface named 'Novest' which can estimate the finger position and classify the finger state (out-of-range/hovering/touching) on the back of a hand with a small ranging sensor array attached on the side of a smartwatch, each of the small ranging sensors gets distance data and strength of signal data. With a prototype of Novest, we conducted the first evaluation to assess the basic performance. In this paper, we conducted the formal user study with two different conditions: the sitting/standing conditions to evaluate the practical performance of Novest as an input interface. The results show that the finger position estimation accuracies in the sitting and standing conditions are 4.2 mm and 5.0 mm, respectively. Additionally, the finger state classification accuracies with parameters adjusted by grid search in the sitting and standing conditions are 97.2% and 97.9%, respectively.