AUTHORS: Tomasz Hachaj, Marcin Piekarczyk
Download as PDF
ABSTRACT: This paper describes and evaluates advanced aligning and comparison method dedicated for human motion trajectory analysis. It utilizes Dynamic Time Warping approach and can be applied for relatively long (30 seconds or longer) and complex motion paths. In contrary to other human motion analysis techniques we do not have restriction on motion direction, we use only kinematic data and we are able to compare any foot trajectory no matter how many rotations take place during the motion. As the result an algorithm outputs set of vectors along motion path that corresponds to beginning and end positions of footsteps. The left and right foot is analyzed separately. The difference of two motion paths can be expressed in any DTW-based feature namely minimal, maximal, median, mean and normalized DTW based distance. We have evaluated our method on karate kata dataset that contains four types of motion sequences performed by two black belt Shorin-Ryu karate masters with more than 20 years of experience. The evaluation of our method assured us that our approach can be easily applied for aligning and comparison of any other motion class described by two dimensional motion trajectories. The method can be applied for example in sport or physical therapy exercises data evaluation and it is invariant to body proportion and motion execution speed.
KEYWORDS: Signal processing, Human motion analysis, Path analysis, Dynamic Time Warping, Karate kata
REFERENCES:
[
1] K. Adistambha, C. H. Ritz, and I. S. Burnett.
Motion classification using dynamic time warping. 2008 IEEE 10th Workshop on Multimedia
Signal Processing, pages 622–627, 2008.
[2] A. Aristidou, D. Cohen-Or, J.K. Hodgins, and
A. Shamir. Self-similarity analysis for motion
capture cleaning. Computer Graphics Forum,
37(2):297–309, 2018.
[3] M. J. Black and A. D. Jepson. Recognizing temporal trajectories using the condensation
algorithm. Proceedings Third IEEE International Conference on Automatic Face and Gesture Recognition, pages 16–21, 1998.
[4] Jaron Blackburn and Eraldo Ribeiro. Human
motion recognition using isomap and dynamic
time warping. Proceedings of the 2Nd Conference on Human Motion: Understanding, Modeling, Capture and Animation, pages 285–298,
2007.
[5] Y. Chen, G. Chen, K. Chen, and B. C. Ooi. Efficient processing of warping time series join of
motion capture data. 2009 IEEE 25th International Conference on Data Engineering, pages
1048–1059, 2009.
[6] O. Cigdem, T. De Laet, and J. De Schutter.
Classical and subsequence dynamic time warping for recognition of rigid body motion trajectories. 2013 9th Asian Control Conference
(ASCC), pages 1–6, 2013.
[7] A. Gupta, J. He, J. Martinez, J. J. Little, and R. J.
Woodham. Efficient video-based retrieval of human motion with flexible alignment. 2016 IEEE
Winter Conference on Applications of Computer
Vision (WACV), pages 1–9, 2016.
[8] Tomasz Hachaj and Marek Ogiela. Heuristic
method for calculation of human body translation using data from inertial motion capture costume. International Journal of Electrical and
Electronic Engineering & Telecommunications,
7:26–29, 2018.
[9] Tomasz Hachaj, Marek R. Ogiela, and
Katarzyna Koptyra. Application of assistive computer vision methods to oyama karate
techniques recognition. Symmetry, 7:1670–
1698, 2015.
[10] Tomasz Hachaj, Marcin Piekarczyk, and
Marek R Ogiela. R language source code and
example data for this paper, 2018. https:
//github.com/browarsoftware/
DTWHumanMotionPathAnalysis.
[11] Tomasz Hachaj, Marcin Piekarczyk, and
Marek R. Ogiela. Signal processing methods
in human motion path analysis: a use case for
karate kata. EECS 2018 conference proceedings,
in press, 2019.
[12] Eugene Hsu, Marco da Silva, and Jovan
Popovic. Comparing the difference be- ´
tween front-leg and back-leg round-house
kicks attacking movement abilities in taekwondo. In Proceedings of the 2007 ACM SIGGRAPH/Eurographics symposium on Computer
animation (SCA ’07). Eurographics Association,
Aire-la-Ville, Switzerland, Switzerland, pages
45–52, 2007.
[13] Mingqin Liu, Xiaoguang Zhang, and Guiyun
Xu. Continuous motion classification and segmentation based on improved dynamic time
warping algorithm. International Journal of
Pattern Recognition and Artificial Intelligence,
32:59–66, 2018.
[14] Marion Morel, Richard Kulpa, Anthony Sorel,
Catherine Achard, and Severine Dubuisson. Au- ´
tomatic and generic evaluation of spatial and
temporal errors in sport motions. In Proceedings of the 11th Joint Conference on Computer
Vision, Imaging and Computer Graphics Theory
and Applications, pages 542–551, 2016.
[15] A.S. Soares and Jr A.L. Apolinario. Real- ´
time 3d gesture recognition using dynamic time
warping and simplification methods. Journal of
WSCG, 25:59–66, 2017.
[16] Adam Switonski, Agnieszka Michalczuk, Henryk Josinski, Andrzej Polanski, and Konrad Wojciechowski. Dynamic time warping in gait
classification of motion capture data. World
Academy of Science, Engineering and Technology International Journal of Computer and Information Engineering, 6:1289–1294, 2012.
[17] Aras Yurtman and Billur Barshan. Automated
evaluation of physical therapy exercises using
multi-template dynamic time warping on wearable sensor signals. Computer Methods and Programs in Biomedicine, 117:189–207, 2014.
[18] F. Zhou and F. De la Torre. Generalized time
warping for multi-modal alignment of human
motion. 2012 IEEE Conference on Computer Vision and Pattern Recognition, pages 1282–1289,
2012.
[19] Feng Zhou and Torre Fernando. Canonical time
warping for alignment of human behavior. Advances in Neural Information Processing Systems 22, pages 2286–2294, 2009.
