Processed Data

The biomechanical analysis of human movement, particularly gait, is crucial in fields such as clinical medicine, sports, and rehabilitation. While traditional motion capture (Mocap) systems are effective, they are often limited by their complexity, high cost, and the unnatural settings they require in terms of the gesture and motion environment. Emerging tools like inertial sensors and mark- erless video-based systems offer greater flexibility but encounter challenges in motion cycle segmentation, as they present kinematic data as time series, adding new difficulties to the analysis. This paper introduces a novel machine learning-based system for automatic gait cycle segmentation using features extracted from two easily measurable lower limb kinematic variables: hip and knee extension angles. The proposed method leverages instantaneous information from these angles for segmentation, ensuring versatility and independence from specific data collection methods. This allows for rapid segmentation and potential implemen-tation on lower-performance processors. Experimental results demonstrate the high accuracy and efficiency of the proposed algorithm segmenting the gait cycle.The F1-score was 0.997. By using readily available hip and knee kinematic data and identifying crucial biomechanical relationships, our method offers a versa-tile and practical solution for motion analysis across various clinical and sports applications.

人体运动生物力学分析，尤其是步态分析，在临床医学、运动科学与康复医学等领域至关重要。尽管传统运动捕捉（Motion Capture，简称Mocap）系统效果可靠，但往往受限于其复杂度高、成本高昂，且所需的手势与运动采集环境并非自然状态。新兴的惯性传感器、无标记视频采集系统等工具虽具备更高灵活性，但在运动周期分割任务中面临挑战：这类系统将运动学数据以时间序列形式输出，为分析工作增添了新的难度。本文提出一种基于机器学习的新型步态周期自动分割系统，该系统从两个易于采集的下肢运动学变量——髋关节伸展角与膝关节伸展角——中提取特征实现分割。所提方法利用上述角度的瞬时信息完成分割，兼具通用性且不依赖特定的数据采集方式，可实现快速分割，并有望在低性能处理器上部署。实验结果表明，所提步态周期分割算法具备高精度与高效率，其F1分数（F1-score）达到0.997。本方法借助易于获取的髋、膝关节运动学数据，并挖掘关键生物力学关联，可为各类临床与运动场景下的运动分析提供通用且实用的解决方案。