【摘要】 针对当前网球发球教学局限于教练的经验评估，成本较高且技术掌握较慢的问题，设计并实现了一种基于姿态识别的网球发球综合评估系统。该系统立足于网球发球的宏观过程特征与微观指标特征，依据关键角序列提取发球关键帧，经线性标准区间量化姿态评估指标，基于动作序列比对进行粗评估，融合层次分析法（AHP）和熵权法（EWM）对关键帧指标进行细评估，计算发球姿态综合得分。将该系统的评估方法与AHP、EWM及全局时间综合评估方法进行对比实验，结果表明，该方法的评估准确性比AHP和EWM分别高出14.1%和14.8%，比全局时间综合评估方法高出6.7%，可为网球发球姿态提供客观的评估结果与科学反馈，辅助发球训练。更多还原

【Abstract】 Aiming at the problems of current tennis serving teaching that is limited to the evaluation of coaches’ experience, high cost, and slow mastery of technology, this paper designs and implements a comprehensive evaluation system for tennis serving based on posture recognition. The system is based on the macro process characteristics and micro index characteristics of tennis serving, extracts the key frames of serving based on key angle sequences, quantizes posture evaluation indicators through linear standard intervals, performs rough evaluation based on motion sequence comparison, and integrates analytic hierarchy process(AHP) and entropy weight method(EWM) to perform detailed evaluation of the key frame indicators, and calculates the comprehensive score of serving posture. Comparative experiments were conducted to compare the evaluation method of this system with AHP, EWM, and global time comprehensive evaluation methods. The results show that the evaluation accuracy of this method is 14.1% and 14.8% higher than AHP and EWM, and 6.7% higher than the global time comprehensive evaluation method.更多还原