期刊文献+

遮挡条件下人体局部运动特征多路径识别方法 预览

Multi-path recognition method of human local motion characteristics under shading condition
在线阅读 下载PDF
收藏 分享 导出
摘要 采用传统方法容易受到噪声影响,无法识别伪特征点,导致人体局部运动特征识别精准度较低,为了解决该问题,提出了遮挡条件下多路径识别方法研究。根据一般识别过程,利用深度信息,计算灰度图像中的光流场,并将矢量依次分解到x、y轴,获取实际运动分量。跟踪实际运动分量最小角点,并对其进行检测,以帧间差信息为基础提取角点,采用前后帧差法消除伪特征点。采用加窗方法对去除伪特征点的人体局部运动特征点进行处理,通过计算标准差、偏度和峰值区分人体前走、后退、跑步、上跳和下蹲局部运动,由此完成多路径识别研究。通过实验对比结果可知,该方法最高识别精准度为98.7%,可以利用到实时行为识别项目研究进程中。 Traditional methods are susceptible to noise and can not recognize pseudo-feature points, which results in low accuracy of human local motion feature recognition. To solve this problem,a multipath recognition method under occlusion conditions is proposed. According to the general recognition process,the optical flow field in gray image is calculated by using depth information,and the vectors are decomposed into x and Y axes in turn to obtain the actual motion components. Tracking the minimum corners of the actual motion component and detecting them,extracting the corners based on the interframe difference information,and eliminating the pseudo-feature points by using the pre-and post-frame difference method. The method of adding windows is used to process the local motion feature points of human body which remove the pseudo feature points. The local motion of human body can be distinguished by calculating standard deviation,skewness and peak value,and then the multi-path recognition research is completed. The experimental results show that the highest recognition accuracy of this method is 98.7%,which can be used in the real-time behavior recognition project research process.
作者 陈永浩 CHEN Yong-hao(Sports Department of Xi 'an Medical University,Xi’an 710021,China)
出处 《电子设计工程》 2019年第15期175-178,183共5页 Electronic Design Engineering
基金 西安医学院2015年青年科研基金项目(2015QN22).
关键词 遮挡条件 人体局部运动 特征 多路径识别 光流场 伪特征点 occlusion condition local motion of human body features multi-path recognition optical flow field pseudo-feature points
作者简介 陈永浩(1985-),男,河南光山人,硕士研究生,讲师。研究方向:体育管理与体育教学。
  • 相关文献

参考文献16

二级参考文献117

  • 1席旭刚,金燕,朱海港,高云园.基于小波包熵和支持向量机的手部肌电信号动作识别[J].中南大学学报:自然科学版,2013,0(S2):150-154. 被引量:3
  • 2陈妍妍,刘发林.Ka频段移动卫星系统的Turbo码仿真性能[J].计算机仿真,2006,23(6):88-90. 被引量:3
  • 3岳峥,杨文革,王晓红.Turbo码在无线通信中的系统级电路仿真[J].计算机仿真,2007,24(3):295-298. 被引量:1
  • 4B Becker, E Ortiz. Evaluation of face recognition techniques for application to Facebook [C]// IEEE International Conference, Automatic Face Gesture Recognition. USA: IEEE, 2008: 1-6. 被引量:1
  • 5S M Pizer, E P Amburn. Adaptive histogram equalization and its variations [J]. Computer Vision Graphics & Image Processing (S0734-189X), 1987, 39(3): 355-368. 被引量:1
  • 6Savvides M, Kumar V. Illumination normalization using logarithm transforms for face authentication[C]// Proceedings IAPR AVBPA. Guildford, UK: Springer Berlin Heidelberg, 2003, 2688: 549-556. 被引量:1
  • 7Neelamma K Patil, Vasudha S, Lokesh R Boregowda. A Novel Method for Illumination Normalization for Performance Improvement of Face Recognition System [C]//2013.International Symposium on Electronic System Design. Singapore: IEEE, 2013: 148-152. 被引量:1
  • 8L E Castillo, L A Cament, F J Galdames, et al. Illumination normalization method using Kolmogorov- Nagumo-based statistics for face recognition [J]. Electronics Letters (S0013-5194), 2014, 50(13): 940-942. 被引量:1
  • 9J Ruiz-del-Solar, J Quinteros. Illumination compensation and normalization in eigenspace-based face recognition: A comparative study of different pre-processing approaches [J]. Pattern Recognition Letter (S0167-8655), 2008, 29(14): 1966-1979. 被引量:1
  • 10Swamp Ku Dandpat, Sukadev Meher. Quality Based Illumination Compensation for Face Recognition[C]// 2011 International Conference on Image Information Processing (ICIIP 2011). Himachal Pradesh, INDIA: IEEE, 2011: 1-4. 被引量:1

共引文献22

投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈