非刚体运动特征提取在武术运动中的应用  

Application of non-rigid motion feature extraction in Chinese Wushu

作　　者：喻龙[1] 单紫徽 席本玉[1] 赵冬 YU Long;SHAN Zi-hui;XI Ben-yu;ZHAO Dong(Department of Physical Education of Xi’an Jiaotong University City College,Xi’an 712039,China;Physical Education Center of Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]西安交通大学城市学院体育教研室,西安712039 [2]西安交通大学体育中心,西安710049

出　　处：《信息技术》2024年第7期104-108,共5页Information Technology

基　　金：2021年陕西省教育科学“十四五”规划课题(SGH2-1Y0411);陕西省自然科学基金(2020JM-146)。

摘　　要：受光照与环境因素的影响,非刚体运动中的图像会存在视差干扰,导致非刚体运动特征提取效果不佳,因此提出一种针对非刚体运动的特征提取方法。对非刚体运动图像进行高斯滤波处理,在高斯差分尺度空间内,采用特征描述算子提取初始特征点。根据特征点提取结果,计算非刚体运动特征提取的代价。引入引导滤波方法对非刚体运动图像进行视差效果优化处理,在最小视差约束下,根据置信度聚合与传播策略进行非刚体运动的特征精确提取。以武术运动作为实验对象,实验结果表明:所提方法的特征点提取精度较高,且视差提取误差率不超过2.0%。Due to the influence of illumination and environmental factors,there will be parallax interference in the image in non-rigid motion,which leads to the poor effect of non-rigid motion feature extraction.Therefore,a feature extraction method for non-rigid motion is proposed.The non-rigid moving image is processed by Gaussian filtering,and the initial feature points are extracted by feature description operator in the Gaussian difference scale space.According to the result of feature point extraction,the cost of non-rigid motion feature extraction is calculated.Guided filtering method is introduced to optimize the parallax effect of non-rigid motion images.Under the constraint of minimum parallax,the features of non-rigid motion are accurately extracted according to the confidence aggregation and propagation strategy.Taking Chinese Wushu as the experimental object,the experiment results show that the feature point extraction accuracy of the proposed method is high,and the parallax extraction error rate does not exceed 2.0%.

关 键 词：特征点提取 非刚体运动 最小视差约束 特征提取 武术运动 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]