基于特征回归的单目深度图无标记人体姿态估计

doi:10.16182/j.issn1004731x.joss.18-0143

系统仿真学报 ›› 2020, Vol. 32 ›› Issue (2): 269-277.doi: 10.16182/j.issn1004731x.joss.18-0143

• 仿真模型/系统置信度评估技术 • 上一篇下一篇

基于特征回归的单目深度图无标记人体姿态估计

陈莹, 沈栎

江南大学轻工过程先进控制教育部重点实验室,江苏无锡 214000

收稿日期:2018-03-15 修回日期:2018-07-03 出版日期:2020-02-18 发布日期:2020-02-19
作者简介:陈莹(1976-),女,浙江丽水,博士,教授,研究方向为图像处理、信息融合、模式识别;沈栎(1993-),男,苗族,贵州黔南,硕士生,研究方向为姿态估计。
基金资助:
国家自然科学基金(61573168)

Monocular Depth Image Mark-less Pose Estimation Based on Feature Regression

Chen Ying, Shen Li

Key Laboratory of Advanced Control Light Process, Jiangnan University, Wuxi 214000, China

Received:2018-03-15 Revised:2018-07-03 Online:2020-02-18 Published:2020-02-19

摘要/Abstract

摘要： 单目深度图无标记人体姿态估计问题,由于动作的多样性,人体自遮挡,运动无规律等因素的影响,导致系统准确率低,鲁棒性不强和运行效率低。为此提出一种基于单目深度图点云的特征提取方法和回归方法,利用特征回归和关节点分类,可以在不使用时间信息的情况下,从单目深度图出估计出人体的关节点坐标。实验结果表明,与其他基于单目深度数据的姿态估计方法,以及相同情况下的多目方法比较,该方法的都能保持很好的精度。

关键词: 计算机视觉, 机器学习, 像素分类, 深度图像, 人体姿态估计, 点云

Abstract: Monocular camera mark-less pose estimation system suffers low accuracy, robustness and efficiency due to variety of action, self-occlusion of human body. A method of feature exaction from point clouds was proposed, in which a single-to-multiple (S2M) feature regressor and a joint position regressor were designed to quickly and accurately predict the 3D positions of body joints from a single depth image without any temporal information. Experiment result shows that the estimation accuracy is superior to that of state-of-the-arts and multi-camera based methods.

Key words: computer vision, machine learning, pixel classification, depth image, pose estimation, point clouds

中图分类号:

TP391

陈莹, 沈栎. 基于特征回归的单目深度图无标记人体姿态估计[J]. 系统仿真学报, 2020, 32(2): 269-277.

Chen Ying, Shen Li. Monocular Depth Image Mark-less Pose Estimation Based on Feature Regression[J]. Journal of System Simulation, 2020, 32(2): 269-277.

参考文献

[1] Knoop S, Vacek S, Dillmann R.Sensor fusion for 3D human body tracking with an articulated 3D body model[C]// IEEE International Conference on Robotics and Automation. Orlando, USA: IEEE, 2006: 1686-1691.
[2] Zhu Y, Fujimura K.A Bayesian Framework for Human Body Pose Tracking from Depth Image Sequences[J]. Sensors (S1424-8220), 2010, 10(5): 5280.
[3] Shotton J, Fitzgibbon A, Cook M, et al.Real-Time Human Pose Recognition in Parts from Single Depth Images[C]. IEEE Conference on Computer Vision and Pattern Recognition. Colorado Springs, USA: IEEE, 2011: 1297-1304.
[4] Sharp T.Implementing Decision Trees and Forests on a GPU[C]. Computer Vision - ECCV 2008, European Conference on Computer Vision. Marseille, France: Springer, October 12-18, 2008, Proceedings, 2008: 595-608.
[5] Shepherd B A.An appraisal of a decision tree approach to image classification[C]. Eighth International Joint Conference on Artificial Intelligence. Karlsruhe, Germany: AAAI Press, 1983: 473-475.
[6] Quinlan J R.Induction of Decision Trees[M]. Dordrecht: Kluwer Academic Publishers, 1986: 81-106.
[7] Ye M, Yang R, Pollefeys M.Accurate 3D pose estimation from a single depth image[C]. International Conference on Computer Vision. Barcelona, Spain: IEEE, 2011: 731-738.
[8] Azrour S, Pierard S, Geurts P, et al.A two-step methodology for human pose estimation increasing the accuracy and reducing the amount of learning samples dramatically[C]. International Conference on Advanced Concepts for Intelligent Vision Systems. Antwerp, Belgium. Springer, 2017: 3-14.
[9] 吕东岳, 黄志蓓, 陶冠宏, 等. 使用简易深度成像设备的高尔夫挥杆动态贝叶斯网络三维重建[J]. 电子与信息学报, 2015, 37(9): 2076-2081.
Lü Dongyue, Huang Zhipei, Tao Guanhong, et al.Dynamic Bayesian network model based golf swing 3D reconstruction using simple depth imaging device[J]. Journal of Electronics & Information Technology, 2015, 37(9): 2076-2081.
[10] 吕东岳. 基于简易深度成像设备的个体运动行为重建[D]. 北京: 中国科学院大学, 2015.
Lü Dongyue.Individual motion behavior reconstruction based on simple depth imaging device[D]. Beijing: University of Chinese Academy of Sciences, 2015.
[11] 朱珏钰, 曹亚微, 周书仁, 等. 基于随机森林深度特征选择的人体姿态估计[J]. 计算机工程与应用, 2017, 53(2): 172-176.
Zhu Jueyu, Cao Yawei, Zhou Shuren, et al.Human pose estimation based on random forest depth feature selection[J]. Computer Engineering and Applications, 2017, 53(2): 172-176.
[12] 张晓丽, 姚俊峰, 黄萍. 基于Kinect的实时360度虚拟试衣[J]. 系统仿真学报, 2016, 28(10): 2378-2384.
Zhang Xiaoli, Yao Junfeng, Huang Ping.360-Degree virtual fitting based on Kinect[J]. Journal of System Simulation, 2016, 28(10): 2378-2384.
[13] 孙金虎, 周来水, 安鲁陵. 点云模型法矢调整优化算法[J]. 中国图象图形学报, 2013, 18(7): 844-851.
Sun Jinhu,Zhou Laishui,An Luling.Optimal algorithm for normal adjustment of point clouds[J]. Journal of Image and Graphics, 2013, 18(7): 844-851.
[14] Ofli F, Chaudhry R, Kurillo G, et al.Berkeley MHAD: A comprehensive Multimodal Human Action Database[C]. IEEE Workshop on Applications of Computer Vision. Tampa, USA: IEEE, 2013: 53-60.
[15] Shafaei A, Little J J.Real-Time Human Motion Capture with Multiple Depth Cameras[C]. 13th Conference on Computer and Robot Vision, Victoria, Canada: IEEE, 2016: 24-31.
[16] Shotton J, Girshick R, Fitzgibbon A, et al.Efficient human pose estimation from single depth images[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence (S0162-8828), 2013, 35(12): 2821-2840.
[17] Shafaei A, UBC3V[EB/OL]. https://github.com/ashafaei.
[18] Hoerl A, Kennard R.Ridge Regression: Applications to Nonorthogonal Problems[J]. Technometrics (S0040-1706), 1970, 12(1): 69-82.
[19] Vovk V.Kernel Ridge Regression[M]. Berlin, Heidelberg: Springer, 2013: 105-116.
[20] An S, Liu W, Venkatesh S.Fast cross-validation algorithms for least squares support vector machine and kernel ridge regression[J]. Pattern Recognition (S0031-3203), 2007, 40(8): 2154-2162.

基于特征回归的单目深度图无标记人体姿态估计

Monocular Depth Image Mark-less Pose Estimation Based on Feature Regression

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	张琪, 曾俊杰, 许凯, 秦龙, 尹全军. 基于机器学习的计算机生成兵力行为建模研究综述[J]. 系统仿真学报, 2021, 33(2): 280-287.
[2]	毕振波, 张世友, 杨花, 吴远红. 基于浅层机器学习的视频监控船舶检测综述[J]. 系统仿真学报, 2021, 33(12): 2792-2807.
[3]	阴敬方, 朱登明, 石敏, 王兆其. 基于引导对抗网络的人体深度图像修补方法[J]. 系统仿真学报, 2020, 32(7): 1312-1321.
[4]	魏亮, 薛牧遥, 霍炬, 张金杰. 基于ICP算法的非合作目标特征点云配准优化[J]. 系统仿真学报, 2020, 32(12): 2383-2387.
[5]	冯振杰, 樊煜. 混沌时间序列的高斯过程混合模型预测[J]. 系统仿真学报, 2019, 31(7): 1387-1396.
[6]	杨永利, 胡晓峰, 荣明, 殷小静, 王文祥. 基于机器学习的作战体系能力特征指标挖掘[J]. 系统仿真学报, 2019, 31(6): 1048-1054.
[7]	徐连瑞, 张锦明. 基于Kinect的室内环境建模[J]. 系统仿真学报, 2019, 31(12): 2643-2651.
[8]	贾立山, 刘喆, 孙毅. 基于RMBP神经网络的飞机电气故障智能诊断[J]. 系统仿真学报, 2018, 30(9): 3493-3501.
[9]	黄天羽, 郭芸莹. 面向动作捕捉的非线性时间序列预测方法研究[J]. 系统仿真学报, 2018, 30(7): 2808-2815.
[10]	唐超, 张苗辉, 李伟, 曹峰, 王晓峰, 童晓红. 基于深度图像的人体行为识别[J]. 系统仿真学报, 2018, 30(5): 1641-1649.
[11]	李琦, 户杏启, 赵建敏. 基于机器学习的热源总供热量优化控制[J]. 系统仿真学报, 2018, 30(3): 1134-1143.
[12]	张艳国, 李擎. 基于惯性测量单元的激光雷达点云融合方法[J]. 系统仿真学报, 2018, 30(11): 4334-4339.
[13]	张航, 陈彬, 薛含章, 朱正秋, 王戎骁. 基于无人机和LIDAR的三维场景建模研究[J]. 系统仿真学报, 2017, 29(9): 1914-1920.
[14]	方志力, 温维亮, 郭新宇, 吴建伟. 基于Kinect的三维玉米植株骨架提取[J]. 系统仿真学报, 2017, 29(3): 524-530.
[15]	温维亮, 郭新宇, 杨涛, 赵德达, 苗腾, 朱宏宇, 董成玉. 玉米果穗点云分割方法研究[J]. 系统仿真学报, 2017, 29(12): 3030-3035.