Golf swing tracking and evaluation using Kinect sensor and particle filter

In this paper, we propose a method to track the grip of a golf swing by the use of a Kinect sensor and particle filter and a method to evaluate the swing trajectory using DP matching. Tracking a golf swing with only a Kinect sensor causes tracking of the grip to be difficult, since swing speed is typically fast, both arms overlap each other, and the grip usually becomes hidden behind the swinger´s head. Furthermore, tracking a golf swing with only a particle filter also causes tracking of the grip to be difficult, since tracking failure occurs when the color of the grip and swinger´s arms is similar. Therefore, by using skeleton information output by a Kinect sensor, we divided the poses of a golf swing into four stages. For each stage, we tracked the exact grip position by changing the tracking method. We also evaluated the swing trajectory using DP matching.