Accuracy evaluation of hand motion measurement using 3D range image sensor

Abstract

Techniques capable of measuring hand motion are expected to be applied in various situations. Examples are the practice support systems used in activities such as sport and dance, for instance, the system employed in baseball pitching form analysis. This analysis evaluates the state of the throwing form numerically and visually by measuring the velocity and trajectory of the hand. This requires measurement of the hand by a sensor and is accomplished by using a motion capture system as the method for measuring hand motion with high accuracy. However, it is time consuming to prepare the system. Therefore, a 3D range image sensor, which is inexpensive and easy to prepare, is proposed. However, as the actual measurement accuracy of this sensor is unknown, it was evaluated by calculating the error with the measured value of motion capture as the true value. In this research, we calculated the error between the 3D coordinates and velocity to evaluate the error. As a result, the effectiveness of the measurement accuracy of the hand motion by the 3D range image sensor was confirmed. The results suggest that the 3D range image sensor can be used in various situations involving human activity.