摘要
对三维运动视觉图像目标特征进行提取,在体育训练指导方面具有实际使用意义。在运动视觉图像在采集过程中,由于受到运动员体状、光照,动作复杂度等因素影响,使得采集图像的关键帧发生动态变化。传统的图像目标特征提取方法主要通过进行大量迭代计算筛选关键技术帧,获取所需信息和边缘特征,对关键帧发生动态变化无具体处理,导致目标特征提取不准确。提出采用视点跟踪切换的三维运动视觉图像目标特征提取方法,对三维运动视觉图像目标特征信息进行采集和边缘特征提取,采用图像分割方法对运动目标特征进行三维图像重构,引入视点跟踪切换方法进行阈值分析和角点标定,实现对运动目标特征的准确跟踪和提取。仿真结果表明,采用改进的方法进行特征提取,准确度及效率具有提高,且稳定性较高。
It is meaningful for practical application to extract target feature of three-dimensional moving visual im- age in instructing training. Factors influence the key frame during process of collecting visual image, such as athlete body shape, light and movement complexity. It makes key frame vary dynamically. Traditional extraction method screens key frame to obtain required information and edge feature through massive iterative computation. It does not process the dynamic variation specifically. So it results in poor extraction precision. In this paper, we propose a target feature extraction method to collect feature information and extract edge feature using viewpoint track switching. Then we use image segmentation method to reconstruct three-dimensional image of moving target feature. Viewpoint tracking switching is introduced to analyze threshold and demarcate corner. Finally we achieve precise tracking and extraction of feature. The simulation results show that the modified method can improve the precision and efficiency. It also has high stability.
出处
《计算机仿真》
北大核心
2017年第1期216-219,共4页
Computer Simulation
关键词
三维运动视觉图像
目标特征提取
视点跟踪切换
Three dimensional motion visual images
Target feature extraction
Viewpoint tracking switch
作者简介
乔良（1983-），男（汉族），河南南阳人，硕士研究生，助教，主要研究领域为体育教学及体育旅游。
