A Low-Cost Skiing Motion Capture System Based on Monocular RGB Camera and MINS Fusion

Abstract

Ice and snow sports have entered a rapid development period since the Winter Olympics, and the emergence of indoor simulated skiing machines has driven the indoor skiing market developed in a high-way. However, there are some shortcomings in the use of existing simulated skiing machines: Simulated skiing machines lack a sense of immersion in relatively dry sports. Fortunately, the visual and inertial motion capture technology has developed rapidly, and is used in entertainment, games, film and television widely. Actually, the multi-node based inertial motion capture system makes it inconvenient to wear as many nodes, and monocular human motion capture system is also affected by occlusion. This paper proposes a real-time human body motion capture method by integrating the monocular vision and the inertial sensors. Specifically, the inertial sensors based Micro Inertial Navigation System (MINS) and the RGB cameras based computer vision system are integrated to capture the trajectory and posture of the key nodes (head, arms, trunk, legs, etc.) of skis and skiers in real-time. Moreover, the filtering technology is also utilized to reduce the jitter of visual motion capture data to improve the stability of the overall system. Finally, an indoor human motion capture experiment was conducted to verify the effectiveness of the scheme. The experimental results shown that there is a large error in visual motion capture under occlusion, and the inertial sensor could be integrated to replace the visual motion capture data for more accurate motion capture under occlusion.