基於單個深度攝像機的自行車運動分析軟體之開發

Abstract

近年來，隨著人們節能減碳和健康樂活理念的提升，自行車的使用率節節攀升。自行車具有交通運輸、休閒以及運動健身等功能，更廣泛應用於下肢傷害的復健，然而不正確的使用也會導致關節肌肉的損耗。為了正確引導自行車使用者的踩踏姿勢以及指導自行車把手及座椅的設計與調整，有必要對腳踏車踩踏做生物力學分析。 本研究利用單個深度攝像頭（KINECT v2），加入全域最優化演算法（GOM）以及卡爾曼濾波（Kalman filter）開發精準可靠的三維動作捕捉系統，配合力學量測設備，建立個人化之下肢生物力學模型來求得踩踏自行車時下肢關節之生物力學變數，並開發一完整易於使用之有用戶介面的動作分析軟體。系統開發完成後利用Vicon紅外立體攝影系統作為標準，以自行車踩踏動作來作為驗證動作，評估本研究開發之系統之可靠性和精準度。 本研究順利完成系統之開發,並利用多圈軌跡疊加與擬合的方法可以在75RPM以內的轉速之下取得與Vicon相接近的關節角度的準確度。

In recent years, with the improvement of people's concept of energy-saving, carbon reduction, and health and well-being, the utilization rate of bicycles has been rising. The bicycle has the functions of transportation, leisure, sports, and fitness, which are also widely used in the rehabilitation of lower limb injuries. However, improper use can also lead to wear down of joints and muscles. In order to correctly guide the pedaling posture of bicycle users and guide the design and adjustment of bicycle handlebar and seat, it is necessary to do biomechanical analysis on pedaling. In this study, the depth camera (KINECT v2) was used, the global optimization method (GOM) and Kalman filter were added to develop an accurate and reliable 3d motion capture system, and the personalized biomechanical model of the lower limb was established in conjunction with the mechanical measurement equipment to obtain the biomechanical variables of the lower limb joints during pedaling. This study successfully completed the development of the system, and the accuracy of the joint Angle close to Vicon can be obtained by using the method of multi-loop trajectory superposition and fitting within the rotation speed of 75RPM.