CAJ | 학술논문

人体动作姿态的识别是当前研究的热点，而对于运动员技术姿态的计算是非常困难的但其应用十分广泛。针对竞技体操现场中运动员的技术姿态识别展开研究，通过监控设备对竞技体操现场进行图像采集，对图像进行关节点标定，利用加权量化矩阵表示技术姿态特征用以解决不同关节点的权重区分度问题，经过权重修正后的关节点相对坐标作为技术姿态特征用以解决标定关节点时产生的误差所造成的识别精度下降的问题，基于多种姿态特征建立技术姿态描述算子并构建特征数据库，用于训练基于支持向量机的多类分类器，以实现竞技体操现场中对体操运动员技术姿态特征的识别。实验表明，提出的方法实现了11种体操运动员运动姿态的特征识别，在识别效率和精度上具有令人满意的结果。
인체동작자태적식별시당전연구적열점，이대우운동원기술자태적계산시비상곤난적단기응용십분엄범。침대경기체조현장중운동원적기술자태식별전개연구，통과감공설비대경기체조현장진행도상채집，대도상진행관절점표정，이용가권양화구진표시기술자태특정용이해결불동관절점적권중구분도문제，경과권중수정후적관절점상대좌표작위기술자태특정용이해결표정관절점시산생적오차소조성적식별정도하강적문제，기우다충자태특정건립기술자태묘술산자병구건특정수거고，용우훈련기우지지향량궤적다류분류기，이실현경기체조현장중대체조운동원기술자태특정적식별。실험표명，제출적방법실현료11충체조운동원운동자태적특정식별，재식별효솔화정도상구유령인만의적결과。
Human motion gesture recognition is a hotspot of current research, and for the attitude of the athletes technical calculation is very difficult but its application is very extensive. For athletics gymnastics field athletes in the study of ges?ture recognition technology, through image acquisition monitoring equipment to the athletics gymnastics field, the key to calibrate the image, using a matrix that quantitative technology gesture feature is used to solve the problem of the weight of different key points to distinguish degrees, after weight correction of the key points of relative coordinates as technical pro?file characteristics produced by the key points to solve calibration error caused by identification of the decline of the preci?sion, based on many features to establish technical gesture description operator and build database, based on support vector machine (SVM) is used to training class classifier, in order to realize the athletics gymnastics in the field of gymnasts tech?nology characteristics of gesture recognition. Experiments show that the proposed method implements the 11 kinds of gym?nasts motion feature recognition, the recognition on the efficiency and accuracy with satisfactory results.