基于特征融合的下肢骨骼数据网格化提取算法  

Gridding Extraction Algorithm for Lower Limb Bone Data Based on Feature Fusion

作　　者：霍宇飞 于随然[1] HUO Yu-fei;YU Sui-ran(School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai200240,China)

机构地区：[1]上海交通大学机械工程学院,上海200240

出　　处：《计算机仿真》2021年第8期269-272,281,共5页Computer Simulation

基　　金：国家自然科学基金资助项目(82072042)。

摘　　要：针对现有的下肢骨骼数据网格化提取算法数据处理能力较差,导致骨骼数据提取结果准确率较低等问题,提出基于特征融合的下肢骨骼数据网格化提取算法。根据人体下肢主要关节功能构建下肢运动分析模型,完成下肢骨骼动力学分析。根据分析结果结合节点移动机体积计算公式,得到下肢骨骼运动数据特征。使用网格中值平滑滤波方法对骨骼数据展开处理并设定合适的网格精度,实现下肢骨骼数据网格化提取,完成下肢骨骼数据网格化提取。构建实验环节,通过对比可知,此方法提取结果的准确率较高,数据使用过程中的可靠性较高,具有较高的实用价值。Traditionally,the data processing of the lower limb skeleton data grid extraction algorithm is poor,which leads to the low accuracy of the extraction results.Therefore,this paper proposes an extraction algorithm for lower limb bone data grid based on feature fusion.Through the main joint function of human lower limb,the analysis model of lower limb movement was established to complete the dynamic analysis of lower limb skeleton.Based on the analysis results and calculation formulas,the data features of lower limb bone motion were obtained.The method of median smoothing filter was used to process the bone data and set the appropriate mesh precision,realizing the mesh extraction of lower limb bone data.The results show that the method has high accuracy,reliability and practical value.

关 键 词：骨骼动作识别 特征融合 动力学分析 时变可靠性 

分 类 号：TN911[电子电信—通信与信息系统]